ACS Publications. Most Trusted. Most Cited. Most Read
Quick View

OR SEARCH CITATIONS

My Activity
Publications
CONTENT TYPES

Figure 1Loading Img
Download Hi-Res ImageDownload to MS-PowerPointCite This:J. Chem. Inf. Model. 2023, 63, 11, 3423-3437
RETURN TO ISSUEPREVComputational Chemis...Computational ChemistryNEXT

Fragment Merging Using a Graph Database Samples Different Catalogue Space than Similarity Search

Cite this: J. Chem. Inf. Model. 2023, 63, 11, 3423–3437
Publication Date (Web):May 25, 2023
https://doi.org/10.1021/acs.jcim.3c00276

Copyright © 2022 The Authors. Published by American Chemical Society. This publication is licensed under

CC-BY 4.0.
  • Open Access

Article Views

2384

Altmetric

-

Citations

-
LEARN ABOUT THESE METRICS
PDF (8 MB)
Get e-Alerts
Supporting Info (1)»
SUBJECTS:

Abstract

High Resolution Image
Download MS PowerPoint Slide

Fragment merging is a promising approach to progressing fragments directly to on-scale potency: each designed compound incorporates the structural motifs of overlapping fragments in a way that ensures compounds recapitulate multiple high-quality interactions. Searching commercial catalogues provides one useful way to quickly and cheaply identify such merges and circumvents the challenge of synthetic accessibility, provided they can be readily identified. Here, we demonstrate that the Fragment Network, a graph database that provides a novel way to explore the chemical space surrounding fragment hits, is well-suited to this challenge. We use an iteration of the database containing >120 million catalogue compounds to find fragment merges for four crystallographic screening campaigns and contrast the results with a traditional fingerprint-based similarity search. The two approaches identify complementary sets of merges that recapitulate the observed fragment–protein interactions but lie in different regions of chemical space. We further show our methodology is an effective route to achieving on-scale potency by retrospective analyses for two different targets; in analyses of public COVID Moonshot and Mycobacterium tuberculosis EthR inhibitors, potential inhibitors with micromolar IC50 values were identified. This work demonstrates the use of the Fragment Network to increase the yield of fragment merges beyond that of a classical catalogue search.

This publication is licensed under

CC-BY 4.0.
  • cc licence
  • by licence

Introduction

ARTICLE SECTIONS
Jump To
Fragment-based drug discovery (FBDD) involves the screening of low-molecular-weight compounds (typically ∼200 Da) against a target of interest, which can subsequently be developed into more potent binders. (1,2) While fragments tend to bind with weak affinity (typically in the high micromolar to low millimolar range), (1) their small size and lower likelihood of steric impediments result in increased likelihood of binding and higher hit rates compared with more traditional high-throughput screens (HTS). (3,4) Other benefits of FBDD include smaller library sizes, the ability to cover a greater proportion of chemical space, and greater control over the optimization of desired chemical properties. (5−8)
Fragment screening using X-ray crystallography is a gold-standard technique as it directly reveals the binding mode of the fragment, providing both confirmation of binding and rich structural data to guide the progression of fragments to become potent and specific binders. (9,10) The throughput of the technique has dramatically improved owing to advances in synchrotrons, automation, and algorithms. (11) Nevertheless, a major bottleneck remains in how to efficiently exploit the information contained in crystal complexes for the rapid structure-guided progression of fragments. There are few efficient, fully automated workflows for this purpose, (12−14) though they clearly hold great potential. For instance, Müller et al. (13) used four crystallographic complexes against protein kinase A as seed fragments for the template-based docking of Enamine REAL compounds and synthesized 93 molecules, the most promising of which demonstrated an inhibition constant (Ki) of 744 nM.
There are three main strategies for fragment progression: growing, which involves the addition of atoms or functional groups that reach new and favorable interactions; linking, which involves the design of a molecular linker to join two fragments that bind to distinct regions of the pocket; and merging, which is used for fragments that bind in partially overlapping space by designing compounds that incorporate substructural features from each. (1) We refer to compounds that maintain exact substructures as “pure merges”.
Fragment merging remains poorly served by in silico approaches, and the published successful fragment merges have been manually designed. (15−22) Not only is this not transferable, but it does not allow the full exploitation of even modest numbers of hits, relying instead on the experience and biases of the medicinal chemist. Only in silico approaches will be able to fully sample the relevant chemical space, though the main challenge lies in recapitulating and maintaining the orientation and interactions of the parent fragments. (23)
In contrast, molecular hybridization, which is conceptually similar to fragment merging but is used in lead optimization, has many computational tools available. Typically, these methods rely on an overlapping substructure to exist between a set of input ligands, both in terms of its chemistry and spatial organization. (24−26) BREED (24) was an early implementation that performs hybridization by overlaying ligands from several protein–ligand structures, identifying matching bonds between the set of ligands and swapping fragments around these matched bonds to generate new hybrid structures. Other hybridization tools use an iterative approach to molecular generation. (27−29) Fragment shuffling (29) aligns several protein–ligand complexes and assigns scores to ligand atoms denoting their contribution to overall binding, after which ligands are fragmented and a selected seed fragment is iteratively grown depending on the calculated fragment scores.
Looking further afield, “impure merges” do not incorporate exact substructures of the parent fragments, but instead incorporate those with similar chemotypes or pharmacophoric properties. This is related to scaffold hopping techniques, which involve the substitution of a molecular core with that of a similar chemotype. For example, Recore (30) replaces fragments in a molecule according to their pharmacophore and arrangement of exit vectors, while CReM (31) is a tool that can mutate, grow or link compounds by selecting fragments to be added that share the correct chemical context.
De novo design using generative models has become increasingly common in the field of molecular optimization, particularly for fragment growing (32−37) and linking. (38−41) However, there are no such existing models for performing fragment merging, partly owing to the lack of data and difficulty in generating synthetic data for training. The main limitation of all de novo approaches is that they tend to propose molecules with poor synthetic accessibility and thus are either impossible or difficult and expensive to make; even state-of-the-art approaches for fragment elaboration struggle to design accessible molecules.
There is a further challenge in progressing fragments into an efficient drug discovery campaign, which is that greater focus needs to be placed on finding readily available compounds to allow elucidation of the structure–activity relationship (SAR) for a target. Commercial catalogues are generally used for this purpose as the compounds are guaranteed to be synthetically feasible and are cheap and easy to acquire. Nevertheless, there are limited examples within the literature of formalized workflows that use catalogue searches to identify and filter compounds. (12,42) Existing methods typically employ search techniques such as substructure and similarity searching, (12) yet these methods can be computationally intensive and fingerprint-based similarity metrics can exhibit bias against fragments (as smaller molecules occupy fewer bits).
In this work, we demonstrate the use of a graph database as a method for finding fragment merges in commercial catalogues, specifically, using an iteration of the Fragment Network. The concept of the Fragment Network was first described by Astex Pharmaceuticals. (43) In that paper, the authors mapped chemical space into a graph database and demonstrated its use for optimizing initial hits against protein kinase B and hepatitis C virus protease–helicase. (43) In this study, we reimplemented the Fragment Network using the RDKit cheminformatics toolbox, (44) loaded it with 120 million commercial catalogue compounds, and developed a pipeline to search for fragment merges, followed by several 2D and 3D filters that prioritize candidates by the likelihood of maintaining the binding pose and interactions of the parent fragments. Unlike other hybridization methods, this approach does not require matching overlapping substructures between an existing set of ligands.
Our Fragment Network searches yield complementary results to classical similarity searches using molecular fingerprints, as each technique identifies filtered compounds for pairs of fragments for which the other technique fails, suggesting that the two techniques should be used in parallel. Specifically, the Fragment Network naturally lends itself both to identifying pure merges and to limiting searches to local chemical space, leading to greater search efficiency. It is thus well-suited to fully exploiting all crystallographic data through large-scale enumeration of all possible pairs of fragment hits. Its suitability for fragment progression was confirmed in two retrospective analyses: comparison against experimental data from the COVID Moonshot (10) identified a known inhibitor and close analogues to other known inhibitors with half-maximal inhibitory concentration (IC50) values within the low-to-medium micromolar range. A known binder and close analogues to known inhibitors were also identified using experimental data against Mycobacterium tuberculosis transcriptional repressor protein EthR.

Materials and Methods

ARTICLE SECTIONS
Jump To

Fragment Network Database

The version of the Fragment Network used for this work was compiled in March 2022 and contains XChem fragment libraries and compounds from the Enamine, MolPort, and Chemspace commercial catalogues, totalling >120 million compounds. The code to generate the database is publicly available at https://github.com/InformaticsMatters/fragmentor.
The network was implemented as described in the original paper: (43) generation of the network involves the decomposition of molecules─represented as nodes─into rings, linkers, and substituents, with the iterative removal of these groups resulting in the enumeration of connected nodes. The network is populated with purchasable compounds that are connected via edges, which represent transformations between nodes. A schematic demonstrating the types of transformation that can be made is shown in Figure S1. The corresponding metadata for nodes and edges, describing features such as the substructure being removed or added when traversing an edge, or, optionally, molecular properties of the molecule node, can allow the tailoring of search queries. The XChem implementation uses Neo4j (45) as the graph database platform, and queries are written in Cypher. This language can enable the creation of sophisticated search queries whereby a user can specify parameters such as the number of hops involved in the query (the number of hops refers to the path distance between nodes in the network), the type of hop to be used in the query path (that is, a contraction or expansion), the substructure(s) involved in the transformation, and the filtering of results based on node properties (for example, heavy atom count).

XChem Data Sets

The four test cases are publicly available crystallographic fragment screens from the XChem facility, summarized in Table 1 and Table S1. These data are available to download from the Fragalysis platform (https://fragalysis.diamond.ac.uk/viewer/react/landing). The specific experiments are described below.
Table 1. Summary of Fragment Screens Used as Test Data
TargetType of binding siteNumber of fragmentsNumber of pairs for merginga
DPP11Allosteric1155
PARP14Active1375
nsp13Active935
MproActive19134
a

Value reflects the number of pairs after removing fragment pairs that are highly similar.

A helicase protein, SARS-CoV-2 nonstructural protein 13 (nsp13), forms part of the replication and transcription machinery of SARS-CoV-2, together with 15 other nonstructural proteins. The function of nsp13 is to catalyze the unwinding of genetic material using energy from nucleotide triphosphate hydrolysis. Two sites were found to have good ligandability in a fragment screen; for the purposes of this work, we focus on developing merges of fragments that bind to the nucleotide site, which is positioned between the 1A and 2A helicase subdomains. (46) Nine overlapping fragment hits that offer good merging opportunities were chosen for testing the pipeline.
The SARS-CoV-2 main protease (Mpro) is involved in processing polyproteins pp1a and pp1ab of SARS-CoV-2, which are necessary for replication and transcription. Mpro is a homodimer of two polypeptides: protomers A and B. Each protomer consists of three domains, and the substrate-binding site, which consists of multiple subsites, is positioned in a cleft between domains I and II, which comprise antiparallel β-barrel structures. (47) Twenty-three noncovalent fragments were found to bind to the active site during crystallographic screening. (48) In this work, we focus on designing merges for 19 of these fragments; the remaining four fragments were used for an independent case study described below.
Human poly(ADP-ribose) polymerase 14 (PARP14) is part of a family of proteins involved in post-translational modification. Shared between the proteins is a highly conserved catalytic domain that binds to NAD+ and transfers negatively charged ADP-ribose to the target protein to be modified. (49) This study focuses on 13 fragment hits found to bind to the catalytic site.
Porphyromonas ginigivalis plays a causative role in the development of periodontal disease. (50) The dipeptidyl peptidase (DPP) enzymes are central to the energy metabolism of the bacterium. (50) Fragment screening found hits that bound to two sites: the active site, to which two fragments bound, and to a potential allosteric site. While the mode of action has yet to be established, 11 fragments were found to bind to the allosteric site; the opportunity for inhibitor development means these 11 hits were selected for merging.

Molecular Fingerprint Calculation

All molecular fingerprints were calculated using the RDKit implementation (44) of the Morgan fingerprint (using 2048 bits and a radius of 2). This fingerprint type is used in all similarity and clustering calculations.

Computational Workflow: Querying

The overall pipeline for querying the database and filtering compounds is illustrated in Figure 1. For all compatible pairs of fragments, we query the Fragment Network as described in the following section. We define two fragments as compatible if they are close in space and are not highly similar to each other. Only pairs for which the distance between the closest pair of atoms is <5 Å are considered. The chosen limit allows the identification of both merges and linkers; this distance threshold can be modified according to which type of elaborated compound to favor. We remove fragments that represent close analogues that bind in an equivalent position to ensure that hybrid molecules are substantially different from the parents. To do so, the maximum common substructure (MCS) between the two fragments is calculated. If up to three heavy atoms in either fragment are not included in the MCS, the RMSD is calculated between the MCS atoms, and fragment pairs are removed, by default, if the RMSD is <2 Å (these parameters can be modified by the user). These pairs typically do not represent useful merges as the resulting compounds do not incorporate unique substructures from each fragment. The total numbers of pairs to undergo querying are shown in Table 1.

Figure 1

Figure 1. Pipeline for identifying fragment merges. Fragment hits from crystallographic fragment screens are used for finding fragment merges. All possible pairs of compounds are enumerated for merging (removing those with high similarity). Both the Fragment Network and similarity search are used to identify fragment merges. The Fragment Network enumerates all possible substructures of one of the fragments in the merge while the other fragment is regarded as the seed fragment. A series of optional hops are made away from the seed fragment (up to a maximum of two), after which an expansion is made by incorporating a substructure from the other fragment. The similarity search finds merges by calculating the Tversky (Tv) similarity against every compound in the database using the Morgan fingerprint (2048 bits and radius 2). The Tversky calculation uses α and β values of 0.7 and 0.3, respectively. All compounds with a mean similarity ≥0.4 are retained. The merges pass through a series of 2D and 3D filters, including pose generation with Fragmenstein, to result in scored poses.

High Resolution Image
Download MS PowerPoint Slide

Search by Fragment Network Query

The Fragment Network query aims to identify close neighbors of one fragment, the seed fragment, that incorporate a substructure of the other fragment in the pair. First, all possible substructures of one of the fragments in the pair are enumerated (this is done by traversing down all edges away from the fragment node that denote contractions and extracting the substructures that are removed). Substructures are filtered for those that contain at least three carbon atoms and do not exist in the other fragment in an overlapping position (considered as >50% overlap in volume). This ensures that the substructure makes a substantial and unique contribution to the final merge. Queries identify all nodes within a specified number of hops away from the seed fragment; the query then attempts to make an additional expansion hop from these nodes in which one of the substructures from the other fragment is incorporated (thus resulting in compounds that contain substructures of both original fragments).
To avoid redundancy in queries, for a single fragment, the substructures for all possible paired fragments are enumerated and pooled together (as the same substructure could be present in multiple fragments). The Fragment Network querying process is asymmetric (that is, there are two sets of queries for each pair of fragments, with each fragment undergoing expansion). The number of optional hops is a tunable parameter; increasing the number of hops will result in a deeper search but will lead to an exponential increase in the time required. In this work, we limit the number of hops to a maximum of two.
All retrieved molecules are filtered for those that contain at least 15 heavy atoms to ensure the compounds represent true merges between the fragments. Query paths (that is, the path taken between the seed fragment node and the merge compounds) are limited to those where the node before expansion contains at least six carbon atoms and are not equivalent to the substructure used in the expansion (this prevents the query from retrieving all nodes connected to the substructure or making expansions from small, ubiquitous nodes that make vast numbers of connections). A limit of 3000 molecules per fragment pair was imposed before filtering to yield numbers comparable to that of the similarity search.
As a consequence of limiting the substructure for expansions to molecules containing at least three carbons, this can result in the Fragment Network not being able to incorporate substituents from the original fragment into the final merge. To combat this, we provide an option to perform R-group expansions of the final nodes. This can be performed as part of the initial query or as a postrefinement step of the most promising filtered molecules.

Similarity Search

The similarity search was performed on the equivalent set of purchasable molecules available in the Fragment Network. For each pair of fragments, the Tversky similarity was calculated (using α and β values of 0.7 and 0.3, respectively) against every compound in the database. The Tversky index is an asymmetric metric and thus prioritizes whether the merges replicate the same bits as the fragments. The geometric mean was calculated between the two values, and a mean value of 0.4 was used as a cutoff (the geometric mean was used to penalize compounds that are highly similar to only one of the fragments in the pair). The number of merges per pair was limited to the 3000 with the highest similarity values (Supplementary Figure S2). Merges with <15 heavy atoms were removed to allow fair comparison with merges identified using the Fragment Network. While an alternative approach would be to perform a single query using the union of the fingerprints for the two fragments, we opted against using this approach, as this will result in smaller fragments contributing less weight to the similarity calculation.

Computational Workflow: Filtering

The retrieved merges are passed through a set of 2D and 3D filters to retain the most promising compounds that are most likely to maintain the binding pose and interactions of the parent fragments. The same filters are applied to the merges from both the Fragment Network and similarity searches to allow fair comparison, with the exception of the expansion filter described below, which is only applied to Fragment Network-derived compounds. These filters are designed for large-scale fragment screens and are thus intentionally stringent. The pipeline has been implemented to enable flexibility: in cases where the filtering pipeline may result in few filtered compounds, it is straightforward to modify the filtering pipeline by either removing filtering steps or modifying thresholds, so that the pipeline can be used in all types of situations.

Filtering Using Calculated Molecular Descriptors

Compounds are filtered according to calculated molecular descriptors using Lipinski’s rule of five (51) and a maximum rotatable bond count of ten. (52)
Furthermore, a limit on the number of consecutive nonring bonds was imposed to remove “stringy”, flexible molecules. Thresholds of eight bonds for “linkers” (substructures connecting two rings) and six bonds for “side chains” (substructures connected to only one ring) were imposed. This was done by calculating the maximum path length of the nonring substructures. Thresholds were set by analyzing the number of consecutive nonring bonds in ChEMBL29 (53) drug molecules (after applying the Lipinski and rotatable bond filters) and calculating the 95th percentile (Figure S3).

Filtering Compounds That Resemble Expansions of One Fragment

A filter was implemented to remove compounds that resemble “expansions” (that is, compounds that represent expansions of one fragment rather than true merges). This occurs when the fragment contributing the substructure for expansion does not contribute anything unique to the final merge. First, the MCS between the seed fragment and the merge is calculated and removed from the merge. The MCS is then calculated between the remainder of the molecule and the substructure used for expansion. If the MCS comprises at least three heavy atoms, the merge passes the filter. This filter is only applied to compounds derived from the Fragment Network approach as the search is found to result in a high proportion of these compounds when the substructure used for expansion is already present in the seed fragment.

Filtering Compounds Using Constrained Embedding

This filter assesses whether it is possible to generate physically reasonable conformations using distance-based constraints based on the poses of the parent fragments. The filter retrieves the atomic coordinates of substructures that were derived from the parent fragments by calculating the MCS or using the substructure that was used for expansion in the original query. The two sets of coordinates from both fragments are combined and used as a template for embedding the merge. As there may be multiple possible sets of atomic coordinates (for example, if there are several substructure matches between the MCS and the merge), every possible pair of coordinates is tried with every possible match with the merge. Merges for which it is possible to generate a physically reasonable structure using the RDKit constrained embedding implementation (44) (that is, the bond distances fulfill the limits set by the distance bonds matrix) undergo an energy calculation using the conformation generated by the constrained embedding; the energy is calculated using RDKit (44) using the UFF molecular force field (this force field is used for consistency with the RDKit constrained embedding implementation).
The energy of the constrained conformation is compared with the mean energy of 50 unconstrained conformations to rule out molecules with energetically infeasible poses. The conformations are randomly generated using RDKit and optimized using the UFF force field; the energy is calculated in the same way as the constrained conformations. A total of 50 conformations was deemed to be sufficient according to the rotatable bond count of the unfiltered molecules (the unfiltered data sets show an average number of rotatable bonds of less than 7). (54) If the ratio between the two energy calculations is >7, the molecule does not pass the filter. This threshold was selected based on an equivalent analysis of the PDBbind 2020 data set, (55) selecting the value for the 95th percentile of the energy ratios (Figure S4). In cases where the filtering pipeline results in few filtered compounds, this filter can be removed to increase the number of possible hits. Conformer generation is instead performed using only Fragmenstein (described below), which requires substantially greater compute time.

Filtering Compounds That Clash with the Protein Pocket

Compounds are filtered according to whether the merge fits the protein pocket. RDKit (44) is used to calculate the protrusion volume between the merge and the protein (the proportion of the volume of the merge that protrudes from the protein). Molecules for which at least 16% of the volume clashes with the protein are removed. This threshold was set using analysis of Mpro crystallographic data. The clash distance between each compound and all protein structures was calculated, and the value for the 95th percentile was used to set the threshold (Figure S5).

Filtering Compounds Following Pose Generation with Fragmenstein

Poses are generated using a tool called Fragmenstein, (56) which was developed as part of the COVID Moonshot project. Similar to the constrained embedding filter described above, Fragmenstein attempts to generate poses using the coordinates of the parent fragments, but is more computationally expensive and thus is used at the final stage of filtering. Fragmenstein generates poses by calculating the MCS and the positional overlap between the merge and the fragments and uses the crystallographic atom coordinates for placing the merge into the protein structure. Following placement, the conformation undergoes energy minimization using PyRosetta. (57) Compounds are filtered for those for which Fragmenstein is able to generate a physically reasonable structure using the coordinates of both fragments. Other filters, including a combined RMSD with the fragments of <1 Å, a negative ΔΔG, and the energy filter described above (to rule out unrealistic conformations), are also applied. A timeout was implemented of 10 min per molecule; molecules for which poses were not generated within this time limit were removed (this can occur for up to 10% of compounds entering the filter).

Computational Workflow: Scoring and Analysis

Filtered compounds are subsequently scored using various metrics to allow comparison between the two techniques. Packages used for scoring are described below.

Prediction of Protein–Ligand Interactions

The protein–ligand interaction profiler (PLIP) was used for predicting protein–ligand interactions. PLIP calculates five types of interaction: hydrophobic contacts, π-stacking interactions (face-to-face or edge-to-face), hydrogen bonds (with the protein as a donor or acceptor), salt bridges (with the protein positively or negatively charged and with metal ions), and halogen bonds. (58)
The predicted interactions were compared with those of the parent fragments to discern whether the merge is able to replicate unique interactions made by each of the fragments (representing useful merges). In addition to comparing chemical diversity of the compound sets, we also compared functional diversity by looking at the interactions made by each compound set; the analysis adapts the methodology described in ref (59). Interaction diversity was analyzed for each target by selecting the minimal subset of filtered compounds that represent all possible interactions (compounds are ranked according to the amount of “information” recovered and are added until reaching a final set in which all possible interactions are recovered). This was repeated 100 times, shuffling each time (as multiple compounds will recover equivalent amounts of information). In each subset, the proportion of compounds that was proposed by either the Fragment Network or the similarity search was calculated to evaluate how diverse each compound set is with respect to diversity of interactions made.

Low-Dimensional Projections of Chemical Space

T-SNE plots were used to create low-dimensional representations of the compound data to allow easy visualization of chemical space. (60) The default parameters set in scikit-learn (version 1.0.1) (61) were used to generate the images, with the maximum number of iterations set at 3000.

Retrospective Analysis Using Experimental Data

We provide details of two separate case studies using inhibitors designed against Mpro and M. tuberculosis transcriptional repressor protein EthR. (21)

Mpro

Existing experimental data consisting of IC50 values exist for Mpro and were used for retrospective analysis using a separate test set of fragments. We used data for compounds that were proposed and screened via the COVID Moonshot project and are available from the PostEra website. (10,62) We ran an independent analysis for five fragments that were used as inspiration for the manual design of 24 compounds (under submission name TRY-UNI-714a760b), eight of which have recorded IC50 values in the low-to-medium micromolar range. This analysis was conducted independent of the main analysis, and thus, these five fragments are not included in the main test set described above. Furthermore, one of the inspirational fragments (fragment x1382-0A) is a covalent binder; however, the designed merges do not incorporate the chlorocetamide group and hence provide a useful test scenario for validating our method.

EthR

Nikifirov et al. (21) describe a fragment-merging approach to designing M. tuberculosis EthR inhibitors using two fragment hits that were found to bind to the protein’s hydrophobic cavity in two different locations using X-ray crystallography (denoted compounds 1 and 2). The two fragments were used to propose three fragment merges based on the different arrangements in the cavity (compounds 3–5), of which compounds 4 and 5 were found to overlap with the volumes of their parent fragments. Several analogues of compound 5 were also synthesized (compounds 14–22) and were assayed using surface plasmon resonance (SPR) to give IC50 values. We use compounds 4, 5, and 14–22 to determine whether the Fragment Network is able to identify known binders or close analogues to known binders.

Results

ARTICLE SECTIONS
Jump To
We tested the ability of the Fragment Network to identify merges using the results of crystallographic fragment screens against four targets: DPP11, PARP14, nsp13, and Mpro. Fragment hits against each target were selected, and pairs of fragments were enumerated for merging. We contrasted the results with that of a more standard fingerprint-based similarity search against the equivalent database of compounds. The resulting merges identified from both techniques were passed through a filtering pipeline, comprising both 2D and 3D filters, to prioritize the most promising molecules based on properties such as molecular descriptors, the ability to fit the protein pocket, and whether the compound is predicted to bind in such a way that maintains the orientation of the original fragments. Final poses were generated using Fragmenstein, and the resulting molecules were scored to allow comparison of performance.

The Fragment Network Readily Identifies Pure Merges

For all targets, the Fragment Network yielded significant numbers of candidate merges that qualify as pure merges, as they could be placed in close agreement (<1 Å RMSD) with their parent fragments (Table 2). We differentiate between four types of merging opportunities based on the overlap between the parent fragments (Figure S6):
1.

Complete overlap merges occur when most of the volume of one fragment overlaps with the other.

2.

Partial overlap by ring occurs when only a ring overlaps between fragment pairs. These are the “classical” merges described in the literature and are akin to the ligand overlap required for molecular hybridization. In these cases, the merging process is relatively straightforward as there is a clear hypothesis for the connectivity of the final molecule.

3.

Partial overlap without ring occurs when the overlap is some other substructure. In these cases, the connectivity of the final compound can be nonobvious.

4.

Nonoverlap of parent fragments requires identifying linking opportunities, which were found in the PARP14 and Mpro data sets.

Table 2. Numbers of Filtered Compounds Identified Using a Fragment Network versus Similarity Search
 Number of hitsSearch typeaBefore filteringAfter filtering% filteredNumber of overlapComplete overlap (%)Partial overlap, ring (%)Partial overlap, no ring (%)No overlap (%)
DPP1111FN22,9031980.980.029.470.6NA
  SS85,9192710.3 4.032.064.0NA
PARP1413FN48,320700.100.018.345.136.6
  SS78,116560.1 16.116.137.530.4
nsp139FN36,2395031.410.499.60.0NA
  SS40,1025301.3 1.697.41.0NA
Mpro19FN109,0129520.942.56.849.641.2
  SS169,4249180.5 71.07.112.59.4
a

FN, Fragment Network; SS, similarity search.

Categories 3 and 4 can be considered “linker-like” merges, and for these, it tends to not be immediately evident which pieces of the parent fragments to incorporate into the final merge and how to do so in a way that will result in a physically reasonable structure that maintains the binding pose of the fragments. Nevertheless, for several such pairs, the Fragment Network identified how to join two distinct substructures of the parent fragments by a molecular linker, resulting in chemical series with diversity in the “linker” region, created by the intermediate optional hops in the database query (Figure 2; Figure S7).

Figure 2

Figure 2. The Fragment Network identifies pure merges. (a) A fragment-merging opportunity for the main protease (Mpro) data set. Interactions are predicted using the protein–ligand interaction profiler (PLIP). Hydrogen bonds and π-stacking interactions are shown by cyan and magenta dotted lines, respectively. The PanDDA density is provided for the purple fragment in the Supporting Information (owing to the unusual conformation). (b) The linker-like merge (pose generated using Fragmenstein) joins substructures from partially overlapping fragments by a “linker-like” region, maintaining the hydrogen bond with THR-45; a change in orientation of the thiazole ring (with respect to the thiophene ring in the fragment) enables an additional π-stacking interaction with HIS-41. (c) A fragment-linking opportunity for Mpro. (d) The proposed compound maintains a hydrogen bond with PHE-140 and makes an additional bond with SER-144. It is worth noting that the linker group proposed by the Fragment Network is present in thioacetazone, an oral antibiotic. (e) The fragments and merge in a and b in 2D. (f) The fragments and merge in c and d in 2D.

High Resolution Image
Download MS PowerPoint Slide
The Fragment Network enables searches not just for merges but also linkers, by including fragment pairs up to 5 Å apart. Linking both fragments in their entirety is typically difficult, as it can lead to compounds that are neither purchasable nor synthetically accessible. Instead, our search links partial structures of each parent, thereby generating many candidates; for Mpro, it identified 419 compounds that are as much linkers as merges.

Fragment Network and Similarity Searches Find Comparable Numbers of Compounds

The outputs of the Fragment Network and similarity searches are compared in Table 2. The totals show the number of unique compounds found across all pairs of fragments, even though the same candidate compound can be identified across multiple fragment pairs. All instances of each compound are filtered, as the 3D structures of the parent fragments may affect their chances of successful filtering. The results show that the techniques produce comparable numbers of filtered compounds across all targets, while the filtering efficiency differed substantially between targets, from 0.1% for PARP14 compounds to 1.3%–1.4% for nsp13 compounds.
Figure 3 shows the number of filtered compounds per pair using each technique, highlighting that this value is highly dependent on both the fragment pair and the technique used, with some pairs yielding no filtered merges from either technique. It was therefore not meaningful to compare scoring metrics between techniques, as there are few fragment pairs resulting in comparable numbers of filtered compounds for both techniques.

Figure 3

Figure 3. The Fragment Network and similarity searches identify filtered compounds for different fragment pairs. The numbers of filtered compounds for each fragment pair found using the Fragment Network (blue) or similarity search (orange) are shown across targets (a) dipeptidyl peptidase 11 (DPP11), (b) poly(ADP-ribose) polymerase 14, (PARP14), (c) nonstructural protein 13 (nsp13), and (d) main protease (Mpro). Only pairs that resulted in filtered compounds are shown. Pairs are ordered from right to left according to the number of Fragment Network compounds found. The data show that each search technique was able to identify filtered compounds for pairs where the other technique identified none.

High Resolution Image
Download MS PowerPoint Slide
The number of merges found by the Fragment Network depends on several factors. First, highly connected seed fragments (typically smaller molecules that decompose into simple, common substructures) will result in more possible query paths and thus greater likelihood of finding merges that incorporate the substructure of interest. Second, a query path that involves an ubiquitous, promiscuous node (for example, if an optional hop yields a benzene) will pull out more paths, as these nodes typically make hundreds of thousands of connections. Third, during filtering, if the fragments in the pair are well-aligned and share overlapping substructures, placement of the merge is more likely to be successful. Fourth, if the query molecule is expanded by a common substructure, this will result in many paths. Specifically, the most common substructure used across all targets for expansion was a benzene ring, which is also a substructure for many of the catalogue compounds. For nsp13, 495 of the 510 (97%) filtered compounds (not accounting for redundancy) were identified by incorporating a benzene. However, the substructures incorporated for the other targets were more diverse, with benzene accounting for 37%–52% of expansions (Table S2).
The number of merges identified by the similarity search is affected by different factors; while it identifies comparable numbers of classical merges (category 2; Figure S8) to the Fragment Network, it is much less effective at identifying linkers (category 4), as evident for Mpro, where the similarity searches identify only 90 linker compounds, compared with 419 from the Fragment Network. Completely overlapping (category 1) merges also present a peculiar difficulty, since it is inherently more difficult to filter out compounds that represent expansions rather than true merges, as it is harder to codify the rules for merges that are maintaining bits of the parent fragments rather than substructures. For example, fragments x0195-0A and x0946-0A of the Mpro data set both share a phenylsulfonamide structure (Figure S9), for which the similarity search identified 395 filtered compounds. However, most of the merges were analogues, as they maintained the phenylsulfonamide but had no unique contributions from the remainder of the fragment structures.

The Fragment Network and Similarity Search Identify Compounds from Distinct Areas of Chemical Space

There is very little overlap between the filtered compounds from the two search techniques, with a maximum of eight compounds found to be in common for DPP11 (Table 2). Figure 3 shows that each technique is able to identify merges for merge pairs for which the other technique identifies none, supporting the notion that these techniques are complementary and could be used in parallel to increase the potential resultant hit rate of a catalogue search (Table S3).
The T-SNE projections in Figure 4 further demonstrate that these techniques operate in distinct areas of chemical space, as both techniques seem to be clustered in different regions and show little overlap. This is particularly apparent for nsp13 and Mpro, owing to the greater availability of data. Figure 4 also shows that the Fragment Network-filtered compounds show greater overall chemical diversity. The compounds were clustered using the Taylor–Butina clustering algorithm (using a Tanimoto distance threshold of 0.3), (63) and the numbers of clusters containing only Fragment Network or similarity search compounds were counted. The Fragment Network was found to result in more clusters across all targets (Figure S11). The mean Tversky similarity between all filtered compounds and their parent fragments were also calculated (Figure S12); the Fragment Network compounds are far more dissimilar using these metrics and are able to access areas of chemical space not reached by the fingerprint-based similarity search, resulting in greater diversity in the compounds retrieved.

Figure 4

Figure 4. Fragment Network and similarity search-derived compound sets populate different regions of chemical space. The chemical space occupied by the filtered compound sets is projected into two dimensions using the T-SNE algorithm across targets (a) dipeptidyl peptidase 11 (DPP11), (b) poly(ADP-ribose) polymerase 14 (PARP14), (c) nonstructural protein 13 (nsp13), and (d) main protease (Mpro). Fragment Network compounds are shown in blue, and similarity search compounds are shown in orange. The two compound sets are shown to occupy distinct areas of chemical space.

High Resolution Image
Download MS PowerPoint Slide

Fragment Network and Similarity Searches Identify Compounds That Form Interactions with Residues Not Reached by the Other Technique

Interactions made by the merge compounds were predicted using PLIP. Across both the Fragment Network and similarity searches, the filtered compound sets were found to interact with the same or greater number of residues than the parent fragments used to create the respective merges. Furthermore, the Fragment Network and similarity search filtered compound sets were each found to make interactions with residues that the other compound set was not able to reach, in particular for DPP11 and PARP14; for nsp13, the Fragment Network reached residues not reached by similarity search, while the opposite was shown for Mpro (Figures S13–S16 and Table S4). We also calculated the number of merges that represent true merges in terms of residue-level interactions, meaning merges that are able to replicate unique interactions made by each of the parent fragments (after discounting interactions that are made by both of the parent fragments). The Fragment Network was more efficient for identifying this type of “true merge” for PARP14 and Mpro, while the similarity search performed better for nsp13 and DPP11 (Table S5). This suggests that the two techniques may perform better for different targets.
In order to explore the concept of “interaction diversity”, whereby compounds are selected according to the diversity in the interactions made (functional diversity) rather than their chemical diversity, we performed an analysis to identify the smallest compound sets for each target that make all possible interactions with the protein. This is done across both the Fragment Network and similarity search-identified compounds. In the final compound sets, we found that compounds from both search approaches were represented, again supporting the idea that the two techniques are complementary (Table 3). For the purposes of identifying compounds for purchase and further screening, we argue that these results support performing both techniques in parallel for catalogue search, expanding the set from which to identify the most useful purchase list for further screening and informing SAR.
Table 3. Composition of Functionally Diverse Subsets Identified from Filtered Compound Setsa
TargetFragment Network compounds (%)Similarity search compounds (%)
DPP1139.7 ± 5.060.3 ± 5.0
PARP1440.3 ± 5.459.7 ± 5.4
nsp1374.0 ± 2.826.0 ± 2.8
Mpro30.5 ± 4.369.5 ± 4.3
a

Mean and standard deviation are provided.

Efficiency of the Filtering Pipeline

A series of 2D and 3D filters was implemented to remove compounds that lack desired molecular properties, that may not represent true merges, that do not fit the protein pocket and that are unable to replicate the binding pose of the original fragments. Table S6 shows the percentages of compounds that are removed by each step in the filtering pipeline (as both the percentage of total compounds and the percentage of compounds entering the filter). The Fragment Network search typically identifies larger molecules and molecules that contain long linker regions containing many rotatable or consecutive nonring bonds, while the similarity search compounds are typically more conservative with regard to their size. The constrained embedding filter removes a greater proportion of similarity search compounds compared with Fragment Network compounds. A possible explanation for this is that the Fragment Network is more likely to preserve exact substructures of the parent fragments, and thus, the MCS calculation used to extract the atomic coordinates of atoms to use as the embedding template is more likely to yield a greater number of atoms. Across both techniques, the step involving pose generation with Fragmenstein and filtering based on the resulting conformations is the most restrictive filter. We would expect this step to be the most conservative, as it is the most accurate and computationally intensive filter in our pipeline for determining whether the molecule can adopt a sensible binding pose that mirrors that of the parent fragments.

The Fragment Network Has Efficiency Benefits over Similarity Search

The Fragment Network search is more computationally efficient than the similarity search. The Fragment Network querying was run single-threaded and takes an average of 2–14 min (dependent on the target) per fragment pair. The similarity search was run on 16 CPUs and takes 2–3 min per fragment pair, requiring up to 40 min in total of CPU time. The time required for filtering all merges for a target ranged from 5.2–75.9 CPU days (this value varies considerably depending on the target and search technique). Timings for filtering and placement of molecules using Fragmenstein are provided in Table S7. Conformer generation using Fragmenstein represents the most computationally expensive step; however, this step could easily be replaced with other placement tools such as constrained docking programs to speed up the pipeline if desired.
In comparison, building the Fragment Network database is a more computationally expensive procedure. To perform the fragmentation and processing of five million molecules for input into the Network requires on average 49 CPU days (this value is highly variable and dependent on molecular complexity). However, given the composition of the database and ability to represent entire catalogues, this is a rare operation to undertake as the overall chemical space of the catalogues will not change drastically over short time scales. This precomputation also allows a more thorough exploration of chemical space than possible using a similarity search alone. Moreover, the addition of new molecules (as catalogues are updated) will not require the reprocessing of all molecules and thus would be substantially less intensive. For example, we could perform the fragmentation of approximately 700,000 molecules within 7 CPU days and thus should be able to perform updates at a rate that exceeds the rate of growth of any commercial catalogue.

Retrospective Analysis Using Experimental Data

Mpro

To test whether the Fragment Network is able to identify true binders, a retrospective analysis was performed using data from the public COVID Moonshot. (10) A set of five fragments were originally used to propose the manually designed TRY-UNI-714a760b series of fragment merges, which successfully achieved on-scale potency for several compounds. We queried the Fragment Network with the same fragments and looked for TRY-UNI-714a760b compounds in the results. Three of the five fragments are either highly similar or represent substructures of each other, and thus, pairs between these fragments were removed, resulting in seven pairs of fragments for querying. We also compared our proposed merges against all other compounds with experimental data but without a covalent warhead. This resulted in a total of 1247 compounds with an IC50 value of <99 μM against Mpro recorded by either a fluorescence or RapidFire mass spectrometry (RF-MS) assay.
After filtering, the Fragment Network did identify one known binder, along with compounds highly similar to several other known binders (Figure 5; Figure S17). The direct match, LON-WEI-b2874fec-25, has an RF-MS IC50 value of 59.6 μM; between the filtered compounds and the known binders, there were 13 pairs with Tanimoto similarity of >0.6, including TRY-UNI-714a760b-18 and TRY-UNI-714a760b-22 (Tanimoto similarities of 0.73 and 0.68, respectively; Figure 5d). Where available, the Fragmenstein-generated poses closely mimic the crystal poses. Compound TRY-UNI-714a760b-16 was also identified, which does not have an IC50 value.

Figure 5

Figure 5. The Fragment Network identifies a known binder against Mpro. A Fragment Network search using (a) two fragment hits against the SARS-CoV-2 main protease (Mpro) identifies, (b) a known binder against Mpro (LON-WEI-b2874fec-25; RapidFire mass spectrometry (RF-MS) IC50 value of 59.6 μM) and similar compounds to known binders, (c) JAN-GHE-83b26c96–22 (fluorescence and RF-MS IC50 values of 96.9 μM and 24.5 μM), and (d) TRY-UNI-714a760b-18 (fluorescence and RF-MS IC50 values of 26.2 μM and 13.0 μM). Fragmenstein-predicted merge poses are shown in white, and crystal poses are in cyan. Interactions are predicted using the protein–ligand interaction profiler (PLIP), and key interaction residues are shown. Hydrogen bonds are shown in cyan, and π-stacking interactions are shown in magenta.

High Resolution Image
Download MS PowerPoint Slide
Two additional known binders (JAN-GHE-83b26c96-22 and TRY-UNI-714a760b-18) were retrieved by mitigating the limitation of the current implementation of the Fragment Network described in the section Search by Fragment Network Query, which necessitates expansions to be made with substructures containing at least three carbons. When we performed R-group expansions on the retrieved analogues, using substituents from the original fragments, the additional binders were identified through the addition of a chlorine atom (Figures S18 and 19).
In comparison, the similarity search after filtering identified three of the retrieved analogues (JAN-GHE-83b26c96-22, TRY-UNI-714a760b-18, and TRY-UNI-714a760b-22). However, the similarity search did not identify the Fragment Network match LON-WEI-b2874fec-25, due to the low Tversky similarity between this compound and the parent fragments. This further supports the complementary nature of these two search techniques.

EthR

We perform an additional analysis using two fragments found to bind to EthR (each fragment binds in two different conformations). As the two fragments already exist in the database, the Fragment Network search was performed using the two fragments, and filtering was repeated using all four pairs of the different fragment conformations. This resulted in 10 filtered compounds (an example is shown in Figure S20). While compound 4 was retrieved from the database, it was ruled out during filtering using our stringent default filtering parameters. As compound 4 was removed by the constrained embedding filter (a possible explanation for this is given below), we also tried relaxing the filtering pipeline by only using Fragmenstein for conformer generation, which increased the number of filtered compounds to 148.
Following this change, the Fragment Network was able to successfully identify compound 4 after filtering (Figure 6). This exemplifies the benefit of implementing a flexible filtering pipeline that can be tailored according to the features of the test system. While the compound has weak potency, with an IC50 value of >100 μM, the Fragmenstein-predicted pose shows good overlap with the crystal pose. However, the change in orientation of the five-membered ring between the two structures (and with respect to the original fragment) is likely the reason for the initial failure of the constrained embedding filter, as RDKit was not able to generate a conformation that maintains the orientation of this ring in a physically reasonable way. While the manually designed compound 5 is not present in the network, the Fragment Network identified several compounds that were highly similar to analogues of compound 5, shown in Figure S21, with SPR IC50 values ranging between 2–25 μM. An example is shown in Figure 6d, for which the Fragmenstein-predicted pose again well reflects that of the original crystal pose. Moreover, the two compounds differ by a single nitrogen atom (which is not present in the original fragment and could not be expected to be identified by the Fragment Network pure merge search).

Figure 6

Figure 6. The Fragment Network identifies a known binder against EthR. A Fragment Network search using (a) two fragment hits against (which each bind in two different positions) Mycobacterium tuberculosis transcriptional repressor protein EthR identifies (b) a known binder (compound 4; IC50 value of >100 μM). (c) An alternative crystallographic arrangement of the equivalent fragments identifies (d) a similar compound to a known binder, compound 21 (IC50 value of 22 μM). Fragmenstein-predicted merge poses are shown in white, and crystal poses are in cyan. Hydrogen bonds are shown in cyan.

High Resolution Image
Download MS PowerPoint Slide
Collectively, these results provide validation for the use of the Fragment Network to progress crystallographic fragment hits to compounds with on-scale potencies in multiple systems.

Discussion

ARTICLE SECTIONS
Jump To
We have illustrated the use of the Fragment Network, a graph database containing compounds from commercial catalogues, for finding fragment merges and show that our method can expand the scope of a traditional similarity-based catalogue search and increase the yield of potential follow-ups. We contrast the use of the Fragment Network and a fingerprint-based similarity search to find merges using the results of crystallographic fragment screens. The results support the idea that these two techniques are complementary and should be used in parallel, in that they are able to identify merges for pairs of fragments that are not represented by the other. They show little overlap in the compounds identified and chemical space in which they operate, and they are able to identify compounds that form interactions not represented by the other. In particular, the Fragment Network is well-suited to identifying what we refer to as pure merges, in other words, compounds that incorporate exact substructures of the parent fragments. The Fragment Network exploits a different type of similarity that is able to preserve substructures of the original fragments but that may appear dissimilar using a fingerprint-based metric. This also demonstrates usefulness beyond that of classical molecular hybridization techniques, which require overlapping substructures to exist between the set of input molecules.
The ability of the Fragment Network to identify known inhibitors of Mpro and EthR and highly similar compounds to inhibitors predicted to bind in the same orientation provides initial validation and supports the use of this approach in the FBDD pipeline. In practice, the types of filter used and the thresholds set will be dependent on the application and target. The pipeline has therefore been built with the flexibility to enable users to select which filters and scoring metrics to use and to set desired parameters and thresholds.
The nature of the pure merges identified has some repercussions; while the variation added to the molecule may not necessarily be better for maintaining all possible fragment interactions, we would hope that the preservation of exact substructures may result in the merges being more likely to mimic the original binding pose of the fragments.
Regarding efficiency, one of the main advantages of the Fragment Network approach is the reduced compute time required to run a search. As the size of commercial catalogues is ever-expanding and new approaches are needed for managing the vast availability of compound data, efficient search techniques represent an important avenue of research. Search techniques such as these could become especially relevant for large virtual libraries with increasing coverage of chemical space.
Future work could improve the efficiency of the search by limiting the substructures used in the expansion to those that are responsible for an interaction that we want to preserve in the final merge; similarly, we could ensure that the seed fragment is not reduced down to a substructure that is also not responsible for any interactions. Moreover, we acknowledge that a limitation of the pipeline is the ability to incorporate substituents from the parent fragments due to the limitation imposed on substructures for expansion of containing at least three carbon atoms. Because of the lack of atom mapping between nodes, it is not feasible to perform multiple expansions using more than one substructure from the original fragment while specifying where on the molecule the substituent should be added. Atom mapping is a feature we hope to implement in future iterations of the network, which will allow greater control over where substructures are added to the final merge.

Conclusions

ARTICLE SECTIONS
Jump To
Currently, there are limited available in silico approaches for identifying fragment merges and, in particular, those that identify compounds that are synthetically accessible. We demonstrate the application of a graph database for identifying commercially available fragment merges that will allow rapid follow-up and progression of hits in FBDD campaigns.
Use of the Fragment Network has proved to be an effective way to improve the yield and potential hit rate of a catalogue search and has demonstrated itself to be complementary to a classical search using a fingerprint-based similarity metric. The network was able to successfully identify known inhibitors against Mpro and EthR, the former with micromolar activity. The results reported here support the use of the two techniques in parallel, with the aim of selecting the optimal compound subset for purchase from all enumerated compounds for further assaying and informing the next iteration of synthesis.

Data Availability

ARTICLE SECTIONS
Jump To

The code to run the querying and filtering pipeline is publicly available through https://github.com/oxpig/fragment_network_merges. The code to generate the Fragment Network is available at https://github.com/InformaticsMatters/fragmentor. The query data retrieved from the database searches and the filtered compounds are available from https://zenodo.org/record/7957805. Access to the current snapshot of the database can be provided upon request.

Supporting Information

ARTICLE SECTIONS
Jump To

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.3c00276.

  • Figures and tables referenced in the text (PDF)

Terms & Conditions

Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.

Author Information

ARTICLE SECTIONS
Jump To
  • Corresponding Author
  • Authors
    • Stephanie Wills - Department of Statistics, University of Oxford, Oxford OX1 3LB, United Kingdom;  Centre for Medicines Discovery, University of Oxford, Oxford OX3 7DQ, United Kingdom;  Orcidhttps://orcid.org/0000-0003-4536-1910
    • Ruben Sanchez-Garcia - Department of Statistics, University of Oxford, Oxford OX1 3LB, United Kingdom;  Centre for Medicines Discovery, University of Oxford, Oxford OX3 7DQ, United Kingdom;  Orcidhttps://orcid.org/0000-0001-6156-3542
    • Tim Dudgeon - Informatics Matters, Ltd., Perch Coworking, Franklins House, Bicester OX26 6JU, United Kingdom
    • Stephen D. Roughley - Vernalis (R&D) Limited, Granta Park, Great Abington, Cambridge CB21 6GB, United Kingdom;  Orcidhttps://orcid.org/0000-0003-0683-5693
    • Andy Merritt - LifeArc, Lynton House, 7−12 Tavistock Square, London WC1H 9LT, United Kingdom;  Orcidhttps://orcid.org/0000-0002-0787-1648
    • Roderick E. Hubbard - Vernalis (R&D) Limited, Granta Park, Great Abington, Cambridge CB21 6GB, United Kingdom;  Orcidhttps://orcid.org/0000-0002-8233-7461
    • James Davidson - Vernalis (R&D) Limited, Granta Park, Great Abington, Cambridge CB21 6GB, United Kingdom;  Orcidhttps://orcid.org/0000-0002-8301-1607
    • Frank von Delft - Centre for Medicines Discovery, University of Oxford, Oxford OX3 7DQ, United Kingdom;  Diamond Light Source, Didcot OX11 0DE, United Kingdom;  Research Complex at Harwell, Harwell Science and Innovation Campus, Didcot OX11 0FA, United Kingdom;  Department of Biochemistry, University of Johannesburg, Auckland Park, Johannesburg 2006, South Africa;  Orcidhttps://orcid.org/0000-0003-0378-0017
  • Notes
    The authors declare no competing financial interest.

Acknowledgments

ARTICLE SECTIONS
Jump To

S.W. is supported by the Engineering and Physical Sciences Research Council (EPSRC; Grant No. EP/S024093/1), Vernalis, and LifeArc. This work has been partially funded by Rosetrees Trust [M940]. The authors thank Alpha Lee for helpful discussion and Matteo Ferla for his help with Fragmenstein.

References

ARTICLE SECTIONS
Jump To

This article references 63 other publications.

  1. 1
    Lamoree, B.; Hubbard, R. Current perspectives in fragment-based lead discovery (FBLD). Essays Biochem 2017, 61, 453– 464,  DOI: 10.1042/EBC20170028
    [Crossref], [PubMed], [CAS], Google Scholar
    1
    Current perspectives in fragment-based lead discovery (FBLD)
    Lamoree Bas; Hubbard Roderick E; Hubbard Roderick E
    Essays in biochemistry (2017), 61 (5), 453-464 ISSN:.
    It is over 20 years since the first fragment-based discovery projects were disclosed. The methods are now mature for most 'conventional' targets in drug discovery such as enzymes (kinases and proteases) but there has also been growing success on more challenging targets, such as disruption of protein-protein interactions. The main application is to identify tractable chemical startpoints that non-covalently modulate the activity of a biological molecule. In this essay, we overview current practice in the methods and discuss how they have had an impact in lead discovery - generating a large number of fragment-derived compounds that are in clinical trials and two medicines treating patients. In addition, we discuss some of the more recent applications of the methods in chemical biology - providing chemical tools to investigate biological molecules, mechanisms and systems.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BC1M7pt1Gmtw%253D%253D&md5=35bce3c5978329c6737ffd719135d363
  2. 2
    Davis, B. J.; Roughley, S. D. In Platform Technologies in Drug Discovery and Validation; Goodnow, R. A., Ed.; Annual Reports in Medicinal Chemistry series; Academic Press, 2017; Vol. 50; pp 371– 439.
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  3. 3
    Hann, M. M.; Leach, A. R.; Harper, G. Molecular complexity and its impact on the probability of finding leads for drug discovery. J. Chem. Inf. Model. 2001, 41, 856– 864,  DOI: 10.1021/ci000403i
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    3
    Molecular Complexity and Its Impact on the Probability of Finding Leads for Drug Discovery
    Hann, Michael M.; Leach, Andrew R.; Harper, Gavin
    Journal of Chemical Information and Computer Sciences (2001), 41 (3), 856-864CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
    Using a simple model of ligand-receptor interactions, the interactions between ligands and receptors of varying complexities are studied and the probabilities of binding calcd. It is obsd. that as the systems become more complex the chance of observing a useful interaction for a randomly chosen ligand falls dramatically. The implications of this for the design of combinatorial libraries is explored. A large set of drug leads and optimized compds. is profiled using several different properties relevant to mol. recognition. The changes obsd. for these properties during the drug optimization phase support the hypothesis that less complex mols. are more common starting points for the discovery of drugs. An extreme example of the use of simple mols. for directed screening against thrombin is provided.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXivFynt7w%253D&md5=e89e76c8fb83fc5012369ae30f20e6b7
  4. 4
    Leach, A. R.; Hann, M. M. Molecular complexity and fragment-based drug discovery: ten years on. Curr. Opin. Chem. Biol. 2011, 15, 489– 496,  DOI: 10.1016/j.cbpa.2011.05.008
    [Crossref], [PubMed], [CAS], Google Scholar
    4
    Molecular complexity and fragment-based drug discovery: ten years on
    Leach, Andrew R.; Hann, Michael M.
    Current Opinion in Chemical Biology (2011), 15 (4), 489-496CODEN: COCBF4; ISSN:1367-5931. (Elsevier B.V.)
    We review the concept of mol. complexity in the context of the very simple model of mol. interactions that we introduced over 10 years ago. A summary is presented of efforts to validate this simple model using screening data. The relationship between the complexity model and the problem of sampling chem. space is discussed, together with the relevance of these theor. concepts to fragment-based drug discovery.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXpvFCltrg%253D&md5=19c2d65de80a8bab1059468828da6ce2
  5. 5
    Keserű, G. M.; Erlanson, D. A.; Ferenczy, G. G.; Hann, M. M.; Murray, C. W.; Pickett, S. D. Design principles for fragment libraries: maximizing the value of learnings from pharma fragment-based drug discovery (FBDD) programs for use in academia. J. Med. Chem. 2016, 59, 8189– 8206,  DOI: 10.1021/acs.jmedchem.6b00197
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    5
    Design Principles for Fragment Libraries: Maximizing the Value of Learnings from Pharma Fragment-Based Drug Discovery (FBDD) Programs for Use in Academia
    Keseru, Gyorgy M.; Erlanson, Daniel A.; Ferenczy, Gyorgy G.; Hann, Michael M.; Murray, Christopher W.; Pickett, Stephen D.
    Journal of Medicinal Chemistry (2016), 59 (18), 8189-8206CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Fragment-based drug discovery (FBDD) is well suited for discovering both drug leads and chem. probes of protein function; it can cover broad swaths of chem. space and allows the use of creative chem. FBDD is widely implemented for lead discovery in industry but is sometimes used less systematically in academia. Design principles and implementation approaches for fragment libraries are continually evolving, and the lack of up-to-date guidance may prevent more effective application of FBDD in academia. This Perspective explores many of the theor., practical, and strategic considerations that occur within FBDD programs, including the optimal size, complexity, physicochem. profile, and shape profile of fragments in FBDD libraries, as well as compd. storage, evaluation, and screening technologies. This compilation of industry experience in FBDD will hopefully be useful for those pursuing FBDD in academia.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XmvFSit7s%253D&md5=888fe2cf003c5564c2ec391cf54ea2ac
  6. 6
    Hall, R. J.; Mortenson, P. N.; Murray, C. W. Efficient exploration of chemical space by fragment-based screening. Prog. Biophys. Mol. Biol. 2014, 116, 82– 91,  DOI: 10.1016/j.pbiomolbio.2014.09.007
    [Crossref], [PubMed], [CAS], Google Scholar
    6
    Efficient exploration of chemical space by fragment-based screening
    Hall, Richard J.; Mortenson, Paul N.; Murray, Christopher W.
    Progress in Biophysics & Molecular Biology (2014), 116 (2-3), 82-91CODEN: PBIMAC; ISSN:0079-6107. (Elsevier Ltd.)
    Screening methods seek to sample a vast chem. space in order to identify starting points for further chem. optimization. Fragment based drug discovery exploits the superior sampling of chem. space that can be achieved when the mol. wt. is restricted. Here we show that com. available fragment space is still relatively poorly sampled and argue for highly sensitive screening methods to allow the detection of smaller fragments. We analyze the properties of our fragment library vs. the properties of X-ray hits derived from the library. We particularly consider properties related to the degree of planarity of the fragments.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1ymu7zO&md5=a63e451443207393b781d3d68b10bcd1
  7. 7
    Murray, C. W.; Rees, D. C. The rise of fragment-based drug discovery. Nat. Chem. 2009, 1, 187– 192,  DOI: 10.1038/nchem.217
    [Crossref], [PubMed], [CAS], Google Scholar
    7
    The rise of fragment-based drug discovery
    Murray, Christopher W.; Rees, David C.
    Nature Chemistry (2009), 1 (3), 187-192CODEN: NCAHBB; ISSN:1755-4330. (Nature Publishing Group)
    A review. The search for new drugs is plagued by high attrition rates at all stages in research and development. Chemists have an opportunity to tackle this problem because attrition can be traced back, in part, to the quality of the chem. leads. Fragment-based drug discovery (FBDD) is a new approach, increasingly used in the pharmaceutical industry, for reducing attrition and providing leads for previously intractable biol. targets. FBDD identifies low-mol.-wt. ligands (∼150 Da) that bind to biol. important macromols. The three-dimensional exptl. binding mode of these fragments is detd. using X-ray crystallog. or NMR spectroscopy, and is used to facilitate their optimization into potent mols. with drug-like properties. Compared with high-throughput-screening, the fragment approach requires fewer compds. to be screened, and, despite the lower initial potency of the screening hits, offers more efficient and fruitful optimization campaigns. Here, we review the rise of FBDD, including its application to discovering clin. candidates against targets for which other chem. approaches have struggled.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXmtF2rs78%253D&md5=0bd1a40fd927cce9200f0fa296a7462f
  8. 8
    Roughley, S. D.; Hubbard, R. E. How well can fragments explore accessed chemical space? A case study from heat shock protein 90. J. Med. Chem. 2011, 54, 3989– 4005,  DOI: 10.1021/jm200350g
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    8
    How Well Can Fragments Explore Accessed Chemical Space? A Case Study from Heat Shock Protein 90
    Roughley, Stephen D.; Hubbard, Roderick E.
    Journal of Medicinal Chemistry (2011), 54 (12), 3989-4005CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    A review. This article summerizes the published literature on the generation of inhibitors and design of drug against the mol. chaperone, heat shock protein 90 (Hsp90), as potential therapeutics to treat cancer.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXmsFymurs%253D&md5=90c2c8028cb460c0c6fdc956e7daae28
  9. 9
    Collins, P. M.; Douangamath, A.; Talon, R.; Dias, A.; Brandao-Neto, J.; Krojer, T.; von Delft, F. Achieving a good crystal system for crystallographic X-ray fragment screening. Methods Enzymol 2018, 610, 251– 264,  DOI: 10.1016/bs.mie.2018.09.027
    [Crossref], [PubMed], [CAS], Google Scholar
    9
    Achieving a good crystal system for crystallographic X-ray fragment screening
    Collins, Patrick M.; Douangamath, Alice; Talon, Romain; Dias, Alexandre; Brandao-Neto, Jose; Krojer, Tobias; von Delft, Frank
    Methods in Enzymology (2018), 610 (Modern Approaches in Drug Discovery), 251-264CODEN: MENZAU; ISSN:0076-6879. (Elsevier Inc.)
    A review. The XChem facility at Diamond Light Source offers fragment screening by X-ray crystallog. as a general access user program. The main advantage of X-ray crystallog. as a primary fragment screen is that it yields directly the location and pose of the fragment hits, whether within pockets of interest or merely on surface sites: this is the key information for structure-based design and for enabling synthesis of follow up mols. Extensive streamlining of the screening expt. at XChem has engendered a very active user program that is generating large amts. of data: in 2017,36 academic and industry groups generated 35,000 datasets of uniquely soaked crystals. It has also generated a large no. of learnings concerning the main remaining bottleneck namely, obtaining a suitable crystal system that will support a successful fragment screen. Here we discuss the practicalities of generating screen-ready crystals that have useful electron d. maps, and how to ensure they will be successfully reproduced and usable at a facility outside the home lab.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhsFKrsrvJ&md5=c122d61fb2371eff58429ce90e777c8e
  10. 10
    The COVID Moonshot Consortium. COVID Moonshot: open science discovery of SARS-CoV-2 main protease inhibitors by combining crowdsourcing, high-throughput experiments, computational simulations, and machine learning. bioRxiv Preprint , 2020.  DOI: 10.1101/2020.10.29.339317
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  11. 11
    Douangamath, A.; Powell, A.; Fearon, D.; Collins, P. M.; Talon, R.; Krojer, T.; Skyner, R.; Brandao-Neto, J.; Dunnett, L.; Dias, A.; Aimon, A.; Pearce, N. M.; Wild, C.; Gorrie-Stone, T.; von Delft, F. Achieving efficient fragment screening at XChem facility at Diamond Light Source. JoVE 2021, 171, e62414  DOI: 10.3791/62414
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  12. 12
    Metz, A.; Wollenhaupt, J.; Glöckner, S.; Messini, N.; Huber, S.; Barthel, T.; Merabet, A.; Gerber, H.-D.; Heine, A.; Klebe, G.; Weiss, M. S. Frag4Lead: growing crystallographic fragment hits by catalog using fragment-guided template docking. Acta Crystallogr., Sect. D: Struct. Biol. 2021, 77, 1168– 1182,  DOI: 10.1107/S2059798321008196
    [Crossref], [CAS], Google Scholar
    12
    Frag4Lead: growing crystallographic fragment hits by catalog using fragment-guided template docking
    Metz, Alexander; Wollenhaupt, Jan; Gloeckner, Steffen; Messini, Niki; Huber, Simon; Barthel, Tatjana; Merabet, Ahmed; Gerber, Hans-Dieter; Heine, Andreas; Klebe, Gerhard; Weiss, Manfred S.
    Acta Crystallographica, Section D: Structural Biology (2021), 77 (9), 1168-1182CODEN: ACSDAD; ISSN:2059-7983. (International Union of Crystallography)
    In recent years, crystallog. fragment screening has matured into an almost routine expt. at several modern synchrotron sites. The hits of the screening expt., i.e. small mols. or fragments binding to the target protein, are revealed along with their 3D structural information. Therefore, they can serve as useful starting points for further structure-based hit-to-lead development. However, the progression of fragment hits to tool compds. or even leads is often hampered by a lack of chem. feasibility. As an attractive alternative, compd. analogs that embed the fragment hit structurally may be obtained from com. catalogs. Here, a workflow is reported based on filtering and assessing such potential follow-up compds. by template docking. This means that the crystallog. binding pose was integrated into the docking calcns. as a central starting parameter. Subsequently, the candidates are scored on their interactions within the binding pocket. In an initial proof-of-concept study using five starting fragments known to bind to the aspartic protease endothiapepsin, 28 follow-up compds. were selected using the designed workflow and their binding was assessed by crystallog. Ten of these compds. bound to the active site and five of them showed significantly increased affinity in isothermal titrn. calorimetry of up to single-digit micromolar affinity. Taken together, this strategy is capable of efficiently evolving the initial fragment hits without major synthesis efforts and with full control by X-ray crystallog.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvFegt7%252FF&md5=391b172987e46364a4b877d2a34508fc
  13. 13
    Müller, J.; Klein, R.; Tarkhanova, O.; Gryniukova, A.; Borysko, P.; Merkl, S.; Ruf, M.; Neumann, A.; Gastreich, M.; Moroz, Y. S.; Klebe, G.; Glinca, S. Magnet for the needle in haystack: “Crystal structure first” Fragment hits unlock active chemical matter using targeted exploration of vast chemical spaces. J. Med. Chem. 2022, 65, 15663– 15678,  DOI: 10.1021/acs.jmedchem.2c00813
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    13
    Magnet for the Needle in Haystack: "Crystal Structure First" Fragment Hits Unlock Active Chemical Matter Using Targeted Exploration of Vast Chemical Spaces
    Mueller, Janis; Klein, Raphael; Tarkhanova, Olga; Gryniukova, Anastasiia; Borysko, Petro; Merkl, Stefan; Ruf, Moritz; Neumann, Alexander; Gastreich, Marcus; Moroz, Yurii S.; Klebe, Gerhard; Glinca, Serghei
    Journal of Medicinal Chemistry (2022), 65 (23), 15663-15678CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Fragment-based drug discovery (FBDD) has successfully led to approved therapeutics for challenging and "undruggable" targets. In the context of FBDD, we introduce a novel, multidisciplinary method to identify active mols. from purchasable chem. space. Starting from four small-mol. fragment complexes of protein kinase A (PKA), a template-based docking screen using Enamine's multibillion REAL Space was performed. A total of 93 mols. out of 106 selected compds. were successfully synthesized. Forty compds. were active in at least one validation assay with the most active follow-up having a 13,500-fold gain in affinity. Crystal structures for six of the most promising binders were rapidly obtained, verifying the binding mode. The overall success rate for this novel fragment-to-hit approach was 40%, accomplished in only 9 wk. The results challenge the established fragment prescreening paradigm since the std. industrial filters for fragment hit identification in a thermal shift assay would have missed the initial fragments.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XitlWrtrfI&md5=7d2ae978c7757c8f1b3083fa84c60f9e
  14. 14
    Piticchio, S. G.; Martínez-Cartró, M.; Scaffidi, S.; Rachman, M.; Rodriguez-Arevalo, S.; Sanchez-Arfelis, A.; Escolano, C.; Picaud, S.; Krojer, T.; Filippakopoulos, P.; von Delft, F.; Galdeano, C.; Barril, X. Discovery of novel BRD4 ligand scaffolds by automated navigation of the fragment chemical space. J. Med. Chem. 2021, 64, 17887– 17900,  DOI: 10.1021/acs.jmedchem.1c01108
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    14
    Discovery of Novel BRD4 Ligand Scaffolds by Automated Navigation of the Fragment Chemical Space
    Piticchio, Serena G.; Martinez-Cartro, Miriam; Scaffidi, Salvatore; Rachman, Moira; Rodriguez-Arevalo, Sergio; Sanchez-Arfelis, Ainoa; Escolano, Carmen; Picaud, Sarah; Krojer, Tobias; Filippakopoulos, Panagis; von Delft, Frank; Galdeano, Carles; Barril, Xavier
    Journal of Medicinal Chemistry (2021), 64 (24), 17887-17900CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Fragment-based drug discovery (FBDD) is a very effective hit identification method. However, the evolution of fragment hits into suitable leads remains challenging and largely artisanal. Fragment evolution is often scaffold-centric, meaning that its outcome depends crucially on the chem. structure of the starting fragment. Considering that fragment screening libraries cover only a small proportion of the corresponding chem. space, hits should be seen as probes highlighting privileged areas of the chem. space rather than actual starting points. We have developed an automated computational pipeline to mine the chem. space around any specific fragment hit, rapidly finding analogs that share a common interaction motif but are structurally novel and diverse. On a prospective application on the bromodomain-contg. protein 4 (BRD4), starting from a known fragment, the platform yields active mols. with nonobvious scaffold changes. The procedure is fast and inexpensive and has the potential to uncover many hidden opportunities in FBDD.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXislWrtbjE&md5=98aebd322c14abb1a7f56157e46013a6
  15. 15
    Miyake, Y.; Itoh, Y.; Hatanaka, A.; Suzuma, Y.; Suzuki, M.; Kodama, H.; Arai, Y.; Suzuki, T. Identification of novel lysine demethylase 5-selective inhibitors by inhibitor-based fragment merging strategy. Bioorg. Med. Chem. 2019, 27, 1119– 1129,  DOI: 10.1016/j.bmc.2019.02.006
    [Crossref], [PubMed], [CAS], Google Scholar
    15
    Identification of novel lysine demethylase 5-selective inhibitors by inhibitor-based fragment merging strategy
    Miyake, Yuka; Itoh, Yukihiro; Hatanaka, Atsushi; Suzuma, Yoshinori; Suzuki, Miki; Kodama, Hidehiko; Arai, Yoshinobu; Suzuki, Takayoshi
    Bioorganic & Medicinal Chemistry (2019), 27 (6), 1119-1129CODEN: BMECEP; ISSN:0968-0896. (Elsevier B.V.)
    Histone lysine demethylases (KDMs) have drawn much attention as targets of therapeutic agents. KDM5 proteins, which are Fe(II)/α-ketoglutarate-dependent demethylases, are assocd. with oncogenesis and drug resistance in cancer cells, and KDM5-selective inhibitors are expected to be anticancer drugs. However, few cell-active KDM5 inhibitors have been reported and there is an obvious need to discover more. In this study, we pursued the identification of highly potent and cell-active KDM5-selective inhibitors. Based on the reported KDM5 inhibitors, we designed several compds. by strategically merging two fragments for competitive inhibition with α-ketoglutarate and for KDM5-selective inhibition. Among them, compds. 10 and 13, which have a 3-cyano pyrazolo[1,5-a]pyrimidin-7-one scaffold, exhibited strong KDM5-inhibitory activity and significant KDM5 selectivity. In cellular assays using human lung cancer cell line A549, 10 and 13 increased the levels of trimethylated lysine 4 on histone H3, which is a specific substrate of KDM5s, and induced growth inhibition of A549 cells. These results should provide a basis for the development of cell-active KDM5 inhibitors to highlight the validity of our inhibitor-based fragment merging strategy.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXivVWltbg%253D&md5=1d9473e11de97ee34b82b82b8d6ffb38
  16. 16
    Ren, J.; Li, J.; Wang, Y.; Chen, W.; Shen, A.; Liu, H.; Chen, D.; Cao, D.; Li, Y.; Zhang, N.; Xu, Y.; Geng, M.; He, J.; Xiong, B.; Shen, J. Identification of a new series of potent diphenol HSP90 inhibitors by fragment merging and structure-based optimization. Bioorg. Med. Chem. Lett. 2014, 24, 2525– 2529,  DOI: 10.1016/j.bmcl.2014.03.100
    [Crossref], [PubMed], [CAS], Google Scholar
    16
    Identification of a new series of potent diphenol HSP90 inhibitors by fragment merging and structure-based optimization
    Ren, Jing; Li, Jian; Wang, Yueqin; Chen, Wuyan; Shen, Aijun; Liu, Hongchun; Chen, Danqi; Cao, Danyan; Li, Yanlian; Zhang, Naixia; Xu, Yechun; Geng, Meiyu; He, Jianhua; Xiong, Bing; Shen, Jingkang
    Bioorganic & Medicinal Chemistry Letters (2014), 24 (11), 2525-2529CODEN: BMCLE8; ISSN:0960-894X. (Elsevier B.V.)
    Heat shock protein 90 (HSP90) is a mol. chaperone to fold and maintain the proper conformation of many signaling proteins, esp. some oncogenic proteins and mutated unstable proteins. Inhibition of HSP90 was recognized as an effective approach to simultaneously suppress several aberrant signaling pathways, and therefore it was considered as a novel target for cancer therapy. Here, by integrating several techniques including the fragment-based drug discovery method, fragment merging, computer aided inhibitor optimization, and structure-based drug design, the authors were able to identify a series of HSP90 inhibitors, e.g. I [R1 = H. HO2CCH2, NH2CH2CH2, MeCONHCH2CH2; r2 = H, Br, O2N, etc.; R3 = H, Br, Me, MeO]. Several compds. can inhibit HSP90 with IC50 about 20-40 nM, which is at least 200-fold more potent than initial fragments in the protein binding assay. These new HSP90 inhibitors not only explore interactions with an under-studied subpocket, also offer new chemotypes for the development of novel HSP90 inhibitors as anticancer drugs.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXmslahsrY%253D&md5=057e15a1982cf723de8c7436accf4fff
  17. 17
    Credille, C. V.; Chen, Y.; Cohen, S. M. Fragment-based identification of influenza endonuclease inhibitors. J. Med. Chem. 2016, 59, 6444– 6454,  DOI: 10.1021/acs.jmedchem.6b00628
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    17
    Fragment-Based Identification of Influenza Endonuclease Inhibitors
    Credille, Cy V.; Chen, Yao; Cohen, Seth M.
    Journal of Medicinal Chemistry (2016), 59 (13), 6444-6454CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    The influenza virus is responsible for millions of cases of severe illness annually. Yearly variance in the effectiveness of vaccination, coupled with emerging drug resistance, necessitates the development of new drugs to treat influenza infections. One attractive target is the RNA-dependent RNA polymerase PA subunit. Herein we report the development of inhibitors of influenza PA endonuclease derived from lead compds. identified from a metal-binding pharmacophore (MBP) library screen. Pyromeconic acid and derivs. thereof were found to be potent inhibitors of endonuclease. Guided by modeling and previously reported structural data, several sublibraries of mols. were elaborated from the MBP hits. Structure-activity relationships were established, and more potent mols. were designed and synthesized using fragment growth and fragment merging strategies. This approach ultimately resulted in the development of a lead compd. with an IC50 value of 14 nM, which displayed an EC50 value of 2.1 μM against H1N1 influenza virus in MDCK cells.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XpsF2mtL8%253D&md5=a872083d373a846467c7f2eb5439bed4
  18. 18
    Edink, E.; Rucktooa, P.; Retra, K.; Akdemir, A.; Nahar, T.; Zuiderveld, O.; van Elk, R.; Janssen, E.; van Nierop, P.; van Muijlwijk-Koezen, J.; Smit, A. B.; Sixma, T. K.; Leurs, R.; de Esch, I. J. P. Fragment growing induces conformational changes in acetylcholine-binding protein: a structural and thermodynamic analysis. J. Am. Chem. Soc. 2011, 133, 5363– 5371,  DOI: 10.1021/ja110571r
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    18
    Fragment Growing Induces Conformational Changes in Acetylcholine-Binding Protein: A Structural and Thermodynamic Analysis
    Edink, Ewald; Rucktooa, Prakash; Retra, Kim; Akdemir, Atilla; Nahar, Tariq; Zuiderveld, Obbe; van Elk, Rene; Janssen, Elwin; van Nierop, Pim; van Muijlwijk-Koezen, Jacqueline; Smit, August B.; Sixma, Titia K.; Leurs, Rob; de Esch, Iwan J. P.
    Journal of the American Chemical Society (2011), 133 (14), 5363-5371CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
    Optimization of fragment hits toward high-affinity lead compds. is a crucial aspect of fragment-based drug discovery (FBDD). In the current study, we have successfully optimized a fragment by growing into a ligand-inducible subpocket of the binding site of acetylcholine-binding protein (AChBP). This protein is a sol. homolog of the ligand binding domain (LBD) of Cys-loop receptors. The fragment optimization was monitored with X-ray structures of ligand complexes and systematic thermodn. analyses using surface plasmon resonance (SPR) biosensor anal. and isothermal titrn. calorimetry (ITC). Using site-directed mutagenesis and AChBP from different species, we find that specific changes in thermodn. binding profiles, are indicative of interactions with the ligand-inducible subpocket of AChBP. This study illustrates that thermodn. anal. provides valuable information on ligand binding modes and is complementary to affinity data when guiding rational structure- and fragment-based discovery approaches.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhvFeisrc%253D&md5=413798f2128e309545de77f8e995b565
  19. 19
    Hughes, S. J.; Millan, D. S.; Kilty, I. C.; Lewthwaite, R. A.; Mathias, J. P.; O’Reilly, M. A.; Pannifer, A.; Phelan, A.; Stühmeier, F.; Baldock, D. A.; Brown, D. G. Fragment based discovery of a novel and selective PI3 kinase inhibitor. Bioorg. Med. Chem. Lett. 2011, 21, 6586– 6590,  DOI: 10.1016/j.bmcl.2011.07.117
    [Crossref], [PubMed], [CAS], Google Scholar
    19
    Fragment based discovery of a novel and selective PI3 kinase inhibitor
    Hughes, Samantha J.; Millan, David S.; Kilty, Iain C.; Lewthwaite, Russell A.; Mathias, John P.; O'Reilly, Mark A.; Pannifer, Andrew; Phelan, Anne; Stuehmeier, Frank; Baldock, Darren A.; Brown, David G.
    Bioorganic & Medicinal Chemistry Letters (2011), 21 (21), 6586-6590CODEN: BMCLE8; ISSN:0960-894X. (Elsevier B.V.)
    We report the use of fragment screening and fragment based drug design to develop a PI3γ kinase fragment hit into a lead. Initial fragment hits were discovered by high concn. biochem. screening, followed by a round of virtual screening to identify addnl. ligand efficient fragments. These were developed into potent and ligand efficient lead compds. using structure guided fragment growing and merging strategies. This led to a potent, selective, and cell permeable PI3γ kinase inhibitor, 12 (I), with good metabolic stability that was useful as a preclin. tool compd.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXht12jtrjO&md5=be9322ee0268040ed1da843025df4cde
  20. 20
    Brough, P. A.; Barril, X.; Borgognoni, J.; Chene, P.; Davies, N. G. M.; Davis, B.; Drysdale, M. J.; Dymock, B.; Eccles, S. A.; Garcia-Echeverria, C.; Fromont, C.; Hayes, A.; Hubbard, R. E.; Jordan, A. M.; Jensen, M. R.; Massey, A.; Merrett, A.; Padfield, A.; Parsons, R.; Radimerski, T.; Raynaud, F. I.; Robertson, A.; Roughley, S. D.; Schoepfer, J.; Simmonite, H.; Sharp, S. Y.; Surgenor, A.; Valenti, M.; Walls, S.; Webb, P.; Wood, M.; Workman, P.; Wright, L. Combining hit identification strategies: fragment-based and in silico approaches to orally active 2-aminothieno[2,3-d]pyrimidine inhibitors of the Hsp90 molecular chaperone. J. Med. Chem. 2009, 52, 4794– 4809,  DOI: 10.1021/jm900357y
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    20
    Combining Hit Identification Strategies: Fragment-Based and in Silico Approaches to Orally Active 2-Aminothieno[2,3-d]pyrimidine Inhibitors of the Hsp90 Molecular Chaperone
    Brough, Paul A.; Barril, Xavier; Borgognoni, Jenifer; Chene, Patrick; Davies, Nicholas G. M.; Davis, Ben; Drysdale, Martin J.; Dymock, Brian; Eccles, Suzanne A.; Garcia-Echeverria, Carlos; Fromont, Christophe; Hayes, Angela; Hubbard, Roderick E.; Jordan, Allan M.; Jensen, Michael Rugaard; Massey, Andrew; Merrett, Angela; Padfield, Antony; Parsons, Rachel; Radimerski, Thomas; Raynaud, Florence I.; Robertson, Alan; Roughley, Stephen D.; Schoepfer, Joseph; Simmonite, Heather; Sharp, Swee Y.; Surgenor, Allan; Valenti, Melanie; Walls, Steven; Webb, Paul; Wood, Mike; Workman, Paul; Wright, Lisa
    Journal of Medicinal Chemistry (2009), 52 (15), 4794-4809CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Inhibitors of the Hsp90 mol. chaperone are showing considerable promise as potential mol. therapeutic agents for the treatment of cancer. Here we describe novel 2-aminothieno[2,3-d]pyrimidine ATP competitive Hsp90 inhibitors, which were designed by combining structural elements of distinct low affinity hits generated from fragment-based and in silico screening exercises in concert with structural information from X-ray protein crystallog. Examples from this series have high affinity (IC50 = 50-100 nM) for Hsp90 as measured in a fluorescence polarization (FP) competitive binding assay and are active in human cancer cell lines where they inhibit cell proliferation and exhibit a characteristic profile of depletion of oncogenic proteins and concomitant elevation of Hsp72. Several compds. caused tumor growth regression at well tolerated doses when administered orally in a human BT474 human breast cancer xenograft model.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXos1ehsbs%253D&md5=4d1988060b0f87dc776e1972ea14808e
  21. 21
    Nikiforov, P. O.; Surade, S.; Blaszczyk, M.; Delorme, V.; Brodin, P.; Baulard, A. R.; Blundell, T. L.; Abell, C. A fragment merging approach towards the development of small molecule inhibitors of Mycobacterium tuberculosis EthR for use as ethionamide boosters. Org. Biomol. Chem. 2016, 14, 2318– 2326,  DOI: 10.1039/C5OB02630J
    [Crossref], [PubMed], [CAS], Google Scholar
    21
    A fragment merging approach towards the development of small molecule inhibitors of Mycobacterium tuberculosis EthR for use as ethionamide boosters
    Nikiforov, Petar O.; Surade, Sachin; Blaszczyk, Michal; Delorme, Vincent; Brodin, Priscille; Baulard, Alain R.; Blundell, Tom L.; Abell, Chris
    Organic & Biomolecular Chemistry (2016), 14 (7), 2318-2326CODEN: OBCRAK; ISSN:1477-0520. (Royal Society of Chemistry)
    With the ever-increasing instances of resistance to frontline TB drugs there is the need to develop novel strategies to fight the worldwide TB epidemic. Boosting the effect of the existing second-line antibiotic ethionamide by inhibiting the mycobacterial transcriptional repressor protein EthR is an attractive therapeutic strategy. Herein we report the use of a fragment based drug discovery approach for the structure-guided systematic merging of two fragment mols., each binding twice to the hydrophobic cavity of EthR from M. tuberculosis. These together fill the entire binding pocket of EthR. We elaborated these fragment hits and developed small mol. inhibitors which have a 100-fold improvement of potency in vitro over the initial fragments.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28Xoslajtg%253D%253D&md5=bbe5d5a96e650841b351f2fb9d221f22
  22. 22
    Schade, M.; Merla, B.; Lesch, B.; Wagener, M.; Timmermanns, S.; Pletinckx, K.; Hertrampf, T. Highly selective sub-nanomolar cathepsin S inhibitors by merging fragment binders with nitrile inhibitors. J. Med. Chem. 2020, 63, 11801– 11808,  DOI: 10.1021/acs.jmedchem.0c00949
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    22
    Highly Selective Sub-Nanomolar Cathepsin S Inhibitors by Merging Fragment Binders with Nitrile Inhibitors
    Schade, Markus; Merla, Beatrix; Lesch, Bernhard; Wagener, Markus; Timmermanns, Simone; Pletinckx, Katrien; Hertrampf, Torsten
    Journal of Medicinal Chemistry (2020), 63 (20), 11801-11808CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Pharmacol. inhibition of cathepsin S (CatS) allows for a specific modulation of the adaptive immune system and many major diseases. Here, we used NMR fragment screening and crystal structure-aided merging to synthesize novel, highly selective CatS inhibitors with picomolar enzymic Ki values and nanomolar functional activity in human Raji cells. Noncovalent fragment hits revealed binding hotspots, while the covalent inhibitor structure-activity relationship enabled efficient potency optimization.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhslGqsrnI&md5=f7698389131c792e3bebb40123f6ec66
  23. 23
    de Souza Neto, L. R.; Moreira-Filho, J. T.; Neves, B. J.; Maidana, R. L. B. R.; Guimarães, A. C. R.; Furnham, N.; Andrade, C. H.; Silva, F. P. In silico strategies to support fragment-to-lead optimization in drug discovery. Front. Chem. 2020, 8, 93,  DOI: 10.3389/fchem.2020.00093
    [Crossref], [PubMed], [CAS], Google Scholar
    23
    In silico Strategies to Support Fragment-to-Lead Optimization in Drug Discovery
    de Souza Neto Lauro Ribeiro; Maidana Rocio Lucia Beatriz Riveros; Silva Floriano Paes Jr; Moreira-Filho Jose Teofilo; Neves Bruno Junior; Andrade Carolina Horta; Neves Bruno Junior; Maidana Rocio Lucia Beatriz Riveros; Guimaraes Ana Carolina Ramos; Furnham Nicholas
    Frontiers in chemistry (2020), 8 (), 93 ISSN:2296-2646.
    Fragment-based drug (or lead) discovery (FBDD or FBLD) has developed in the last two decades to become a successful key technology in the pharmaceutical industry for early stage drug discovery and development. The FBDD strategy consists of screening low molecular weight compounds against macromolecular targets (usually proteins) of clinical relevance. These small molecular fragments can bind at one or more sites on the target and act as starting points for the development of lead compounds. In developing the fragments attractive features that can translate into compounds with favorable physical, pharmacokinetics and toxicity (ADMET-absorption, distribution, metabolism, excretion, and toxicity) properties can be integrated. Structure-enabled fragment screening campaigns use a combination of screening by a range of biophysical techniques, such as differential scanning fluorimetry, surface plasmon resonance, and thermophoresis, followed by structural characterization of fragment binding using NMR or X-ray crystallography. Structural characterization is also used in subsequent analysis for growing fragments of selected screening hits. The latest iteration of the FBDD workflow employs a high-throughput methodology of massively parallel screening by X-ray crystallography of individually soaked fragments. In this review we will outline the FBDD strategies and explore a variety of in silico approaches to support the follow-up fragment-to-lead optimization of either: growing, linking, and merging. These fragment expansion strategies include hot spot analysis, druggability prediction, SAR (structure-activity relationships) by catalog methods, application of machine learning/deep learning models for virtual screening and several de novo design methods for proposing synthesizable new compounds. Finally, we will highlight recent case studies in fragment-based drug discovery where in silico methods have successfully contributed to the development of lead compounds.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB387psFOgtg%253D%253D&md5=3fac2ef52a5f51e4e34826f08d5e28d5
  24. 24
    Pierce, A. C.; Rao, G.; Bemis, G. W. BREED:generating novel inhibitors through hybridization of known ligands. Application to CDK2, P38, and HIV protease. J. Med. Chem. 2004, 47, 2768– 2775,  DOI: 10.1021/jm030543u
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    24
    BREED: Generating Novel Inhibitors through Hybridization of Known Ligands. Application to CDK2, P38, and HIV Protease
    Pierce, Albert C.; Rao, Govinda; Bemis, Guy W.
    Journal of Medicinal Chemistry (2004), 47 (11), 2768-2775CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    In this work we describe BREED, a method for the generation of novel inhibitors from structures of known ligands bound to a common target. The method is essentially an automation of the common medicinal chem. practice of joining fragments of two known ligands to generate a new inhibitor. The ligand-bound target structures are overlaid, all overlapping bonds in all pairs of ligands are found, and the fragments on each side of each matching bond are swapped to generate the new mols. Since the method is automated, it can be applied recursively to generate all possible combinations of known ligands. In an application of this method to HIV protease inhibitors and protein kinase inhibitors, hundreds of new mol. structures were generated. These included known inhibitor scaffolds not included in the initial set, entirely novel scaffolds, and novel substituents on known scaffolds. The method is fast, and since all of the ligand functional groups are known to bind the target in the precise position and orientation present in the novel ligand, the success rate of this method should be superior to more traditional de novo design techniques. In an era of increasingly high-throughput structural biol., such methods for high-throughput utilization of structural information will become increasingly valuable.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXjt1ajurk%253D&md5=e0bec9082fab05b694742498b3f41549
  25. 25
    Lindert, S.; Durrant, J. D.; McCammon, J. A. LigMerge: a fast algorithm to generate models of novel potential ligands from sets of known binders. Chem. Biol. Drug Des. 2012, 80, 358– 365,  DOI: 10.1111/j.1747-0285.2012.01414.x
    [Crossref], [PubMed], [CAS], Google Scholar
    25
    LigMerge: a fast algorithm to generate models of novel potential ligands from sets of known binders
    Lindert, Steffen; Durrant, Jacob D.; McCammon, J. Andrew
    Chemical Biology & Drug Design (2012), 80 (3), 358-365CODEN: CBDDAL; ISSN:1747-0277. (Wiley-Blackwell)
    One common practice in drug discovery is to optimize known or suspected ligands to improve binding affinity. In performing these optimizations, it is useful to look at as many known inhibitors as possible for guidance. Medicinal chemists often seek to improve potency by altering certain chem. moieties of known/endogenous ligands while retaining those crit. for binding. To our knowledge, no automated, ligand-based algorithm exists for systematically "swapping" the chem. moieties of known ligands to generate novel ligands with potentially improved potency. To address this need, we have created a novel algorithm called "LigMerge". LigMerge identifies the max. (largest) common substructure of two three-dimensional ligand models, superimposes these two substructures, and then systematically mixes and matches the distinct fragments attached to the common substructure at each common atom, thereby generating multiple compd. models related to the known inhibitors that can be evaluated using computer docking prior to synthesis and exptl. testing. To demonstrate the utility of LigMerge, we identify compds. predicted to inhibit peroxisome proliferator-activated receptor gamma, HIV reverse transcriptase, and dihydrofolate reductase with affinities higher than those of known ligands. We hope that LigMerge will be a helpful tool for the drug design community.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XhtF2itbbF&md5=07cc659c48eccc967fc3b135182e9012
  26. 26
    Wang, H.; Pan, X.; Zhang, Y.; Wang, X.; Xiao, X.; Ji, C. MolHyb: a web server for structure-based drug design by molecular hybridization. J. Chem. Inf. Model. 2022, 62, 2916– 2922,  DOI: 10.1021/acs.jcim.2c00443
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    26
    MolHyb: A Web Server for Structure-Based Drug Design by Molecular Hybridization
    Wang, Hao; Pan, Xiaolin; Zhang, Yueqing; Wang, Xingyu; Xiao, Xudong; Ji, Changge
    Journal of Chemical Information and Modeling (2022), 62 (12), 2916-2922CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Mol. hybridization is a widely used ligand design method in drug discovery. In this study, we present MolHyb, a web server for structure-based ligand design by mol. hybridization. The input of MolHyb is a protein file and a seed compd. file. MolHyb tries to generate novel ligands through hybridizing the seed compd. with helper compds. that bind to the same protein target or similar proteins. To facilitate the job of getting helper compds., we compiled a modeled protein-ligand structure database as an extension to crystal structures in the PDB database by placing the bioactive compds. in ChEMBL into their corresponding 3D protein binding pocket properly. MolHyb works by searching for helper compds. from the protein-ligand structure database and migrating chem. moieties from helper compds. to the seed compd. efficiently. Hybridization is performed at both cyclic and acyclic bonds. The users can also input their own helper compds. to MolHyb. We hope that MolHyb will be a useful tool for rational drug design. MolHyb is freely available at http://molhyb.xundrug.cn/.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhsVyjsr3I&md5=f7682a620c9ca25bfbc1142b204e20fa
  27. 27
    Li, Y.; Zhao, Y.; Liu, Z.; Wang, R. Automatic Tailoring and Transplanting: a practical method that makes virtual screening more useful. J. Chem. Inf. Model. 2011, 51, 1474– 1491,  DOI: 10.1021/ci200036m
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    27
    Automatic Tailoring and Transplanting: A Practical Method that Makes Virtual Screening More Useful
    Li, Yan; Zhao, Yuan; Liu, Zhihai; Wang, Renxiao
    Journal of Chemical Information and Modeling (2011), 51 (6), 1474-1491CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Docking-based virtual screening of large compd. libraries has been widely applied to lead discovery in structure-based drug design. However, subsequent lead optimizations often rely on other types of computational methods, such as de novo design methods. We have developed an automatic method, namely automatic tailoring and transplanting (AutoT&T), which can effectively utilize the outcomes of virtual screening in lead optimization. This method detects suitable fragments on virtual screening hits and then transplants them onto a lead compd. to generate new ligand mols. Binding affinities, synthetic feasibilities, and drug-likeness properties are considered in the selection of final designs. In this study, our AutoT&T program was tested on three different target proteins, including p38 MAP kinase, PPAR-α, and Mcl-1. In the first two cases, AutoT&T was able to produce mols. identical or similar to known inhibitors with better potency than the given lead compd. In the third case, we demonstrated how to apply AutoT&T to design novel ligand mols. from scratch. Compared to the solns. generated by other two de novo design methods, i.e., LUDI and EA-Inventor, the solns. generated by AutoT&T were structurally more diverse and more promising in terms of binding scores in all three cases. AutoT&T also completed the assigned jobs more efficiently than LUDI and EA-Inventor by several folds. Our AutoT&T method has certain tech. advantages over de novo design methods. Importantly, it expands the application of virtual screening from lead discovery to lead optimization and thus may serve as a valuable tool for many researchers.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXlvFKqsb0%253D&md5=3327b42fecda9f2af07a6864d05afaa4
  28. 28
    Li, Y.; Zhao, Z.; Liu, Z.; Su, M.; Wang, R. AutoT&T v.2: an efficient and versatile tool for lead structure generation and optimization. J. Chem. Inf. Model. 2016, 56, 435– 453,  DOI: 10.1021/acs.jcim.5b00691
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    28
    AutoT&T v.2: An Efficient and Versatile Tool for Lead Structure Generation and Optimization
    Li, Yan; Zhao, Zhixiong; Liu, Zhihai; Su, Minyi; Wang, Renxiao
    Journal of Chemical Information and Modeling (2016), 56 (2), 435-453CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    In structure-based drug design, automated de novo design methods are helpful tools for lead discovery as well as lead optimization. In a previous study the authors reported a new de novo design method, namely, Automatic Tailoring and Transplanting (AutoT&T). It overcomes some intrinsic problems in conventional fragment-based buildup methods. In this study, the authors describe an upgraded version, namely, AutoT&T2. Structural operations conducted by AutoT&T2 have been largely optimized by introducing several new algorithms. As a result, its overall speed in multiround optimization jobs has been improved by a few thousand fold. With this improvement, it is now practical to conduct structural crossover among multiple lead mols. using AutoT&T2. Three different test cases are described in this study that demonstrate the new features and versatile applications of AutoT&T2. The AutoT&T2 software suite is available to the public. Besides, a Web portal for running AutoT&T2 online is provided at http://www.sioc-ccbg.ac.cn/software/att2 for testing.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28Xht1KgurY%253D&md5=765cef8f54f2367d902682c547abaf52
  29. 29
    Nisius, B.; Rester, U. Fragment shuffling: an automated workflow for three-dimensional fragment-based ligand design. J. Chem. Inf. Model. 2009, 49, 1211– 1222,  DOI: 10.1021/ci8004572
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    29
    Fragment Shuffling: An Automated Workflow for Three-Dimensional Fragment-Based Ligand Design
    Nisius, Britta; Rester, Ulrich
    Journal of Chemical Information and Modeling (2009), 49 (5), 1211-1222CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Fragment-based approaches display a promising alternative in lead discovery. Herein, we present the automated fragment shuffling workflow for the identification of novel lead compds. combining central elements from fragment-based lead identification and structure-based de novo design. Our method is based on sets of aligned 3D ligand structures binding to the same target or target family. The implementation comprises three different ligand fragmentation methods, a scoring scheme assigning individual scores to each fragment, and the incremental construction of novel ligands based on a greedy search algorithm guided by the calcd. fragment scores. The validation of our 3D ligand design workflow is presented on the basis of two pharmaceutically relevant drug targets. A retrospective study based on a selected protein kinase data set revealed that the fragment shuffling approach realizes extended results compared to the well-known BREED technique. Furthermore, we applied our approach in a prospective study for the design of novel non-peptidic thrombin inhibitors. The designed ligand structures in both studies demonstrate the potential of the fragment shuffling workflow.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXltlKls74%253D&md5=87aae00271c326a921cf38e7770405c8
  30. 30
    Maass, P.; Schulz-Gasch, T.; Stahl, M.; Rarey, M. Recore: a fast and versatile method for scaffold hopping based on small molecule crystal structure conformations. J. Chem. Inf. Model. 2007, 47, 390– 399,  DOI: 10.1021/ci060094h
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    30
    Recore: A Fast and Versatile Method for Scaffold Hopping Based on Small Molecule Crystal Structure Conformations
    Maass, Patrick; Schulz-Gasch, Tanja; Stahl, Martin; Rarey, Matthias
    Journal of Chemical Information and Modeling (2007), 47 (2), 390-399CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Replacing central elements of known active structures is a common procedure to enter new compd. classes. Different computational methods have already been developed to help with this task, varying in the description of possible replacements, the query input, and the similarity measure used. In this paper, a novel approach for scaffold replacement and a corresponding software tool, called Recore, is introduced. In contrast to prior methods, the authors main objective was to combine the following three properties in one tool: to avoid structures with strained conformations, to enable the exploration of large search spaces, and to allow interactive use through short response times. The authors introduce a new technique employing 3D fragments generated by combinatorial enumeration of cuts. It allows focusing on fragments suitable for scaffold replacement while retaining conformational information of the corresponding crystal structures. Based on this idea, the authors present an algorithm utilizing a geometric rank searching approach. Given a geometric arrangement of two or three exit vectors and addnl. pharmacophore features, the algorithm finds fragments fulfilling all these constraints ordered by increasing deviation from the query constraints. For the validation of the approach, three different drug design scenarios have been used. The results obtained show that the authors approach is able to propose new valid scaffold topologies.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2sXhvVWjtbk%253D&md5=e9d4e277da49f4ffe2c21472990c2387
  31. 31
    Polishchuk, P. CReM: chemically reasonable mutations framework for structure generation. J. Cheminf. 2020, 12, 28,  DOI: 10.1186/s13321-020-00431-w
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  32. 32
    Lim, J.; Hwang, S.-Y.; Moon, S.; Kim, S.; Kim, W. Y. Scaffold-based molecular design with a graph generative model. Chem. Sci. 2020, 11, 1153– 1164,  DOI: 10.1039/C9SC04503A
    [Crossref], [CAS], Google Scholar
    32
    Scaffold-based molecular design with a graph generative model
    Lim, Jaechang; Hwang, Sang-Yeon; Moon, Seokhyun; Kim, Seungsu; Kim, Woo Youn
    Chemical Science (2020), 11 (4), 1153-1164CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
    Searching for new mols. in areas like drug discovery often starts from the core structures of known mols. Such a method has called for a strategy of designing deriv. compds. retaining a particular scaffold as a substructure. On this account, our present work proposes a graph generative model that targets its use in scaffold-based mol. design. Our model accepts a mol. scaffold as input and extends it by sequentially adding atoms and bonds. The generated mols. are then guaranteed to contain the scaffold with certainty, and their properties can be controlled by conditioning the generation process on desired properties. The learned rule of extending mols. can well generalize to arbitrary kinds of scaffolds, including those unseen during learning. In the conditional generation of mols., our model can simultaneously control multiple chem. properties despite the search space constrained by fixing the substructure. As a demonstration, we applied our model to designing inhibitors of the epidermal growth factor receptor and show that our model can employ a simple semi-supervised extension to broaden its applicability to situations where only a small amt. of data is available.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXit1Ortr%252FO&md5=be36a65abcce15f18b4c1e529bffd905
  33. 33
    Imrie, F.; Hadfield, T. E.; Bradley, A. R.; Deane, C. M. Deep generative design with 3D pharmacophoric constraints. Chem. Sci. 2021, 12, 14577– 14589,  DOI: 10.1039/D1SC02436A
    [Crossref], [PubMed], [CAS], Google Scholar
    33
    Deep generative design with 3D pharmacophoric constraints
    Imrie, Fergus; Hadfield, Thomas E.; Bradley, Anthony R.; Deane, Charlotte M.
    Chemical Science (2021), 12 (43), 14577-14589CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
    Generative models have increasingly been proposed as a soln. to the mol. design problem. However, it has proved challenging to control the design process or incorporate prior knowledge, limiting their practical use in drug discovery. In particular, generative methods have made limited use of three-dimensional (3D) structural information even though this is crit. to binding. This work describes a method to incorporate such information and demonstrates the benefit of doing so. We combine an existing graph-based deep generative model, DeLinker, with a convolutional neural network to utilize phys.-meaningful 3D representations of mols. and target pharmacophores. We apply our model, DEVELOP, to both linker and R-group design, demonstrating its suitability for both hit-to-lead and lead optimization. The 3D pharmacophoric information results in improved generation and allows greater control of the design process. In multiple large-scale evaluations, we show that including 3D pharmacophoric constraints results in substantial improvements in the quality of generated mols. On a challenging test set derived from PDBbind, our model improves the proportion of generated mols. with high 3D similarity to the original mol. by over 300%. In addn., DEVELOP recovers 10x more of the original mols. compared to the baseline DeLinker method. Our approach is general-purpose, readily modifiable to alternate 3D representations, and can be incorporated into other generative frameworks.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXitlSrur%252FL&md5=1e51c4d1f28c6c8dff140c4c32d817f4
  34. 34
    Hadfield, T. E.; Imrie, F.; Merritt, A.; Birchall, K.; Deane, C. M. Incorporating target-specific pharmacophoric information into deep generative models for fragment elaboration. J. Chem. Inf. Model. 2022, 62, 2280– 2292,  DOI: 10.1021/acs.jcim.1c01311
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    34
    Incorporating Target-Specific Pharmacophoric Information into Deep Generative Models for Fragment Elaboration
    Hadfield, Thomas E.; Imrie, Fergus; Merritt, Andy; Birchall, Kristian; Deane, Charlotte M.
    Journal of Chemical Information and Modeling (2022), 62 (10), 2280-2292CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Despite recent interest in deep generative models for scaffold elaboration, their applicability to fragment-to-lead campaigns has so far been limited. This is primarily due to their inability to account for local protein structure or a user's design hypothesis. We propose a novel method for fragment elaboration, STRIFE, that overcomes these issues. STRIFE takes as input fragment hotspot maps (FHMs) extd. from a protein target and processes them to provide meaningful and interpretable structural information to its generative model, which in turn is able to rapidly generate elaborations with complementary pharmacophores to the protein. In a large-scale evaluation, STRIFE outperforms existing, structure-unaware, fragment elaboration methods in proposing highly ligand-efficient elaborations. In addn. to automatically extg. pharmacophoric information from a protein target's FHM, STRIFE optionally allows the user to specify their own design hypotheses.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhtFGru7rE&md5=c639b01f0c509bbd2e20308fa044bdb5
  35. 35
    Arús-Pous, J.; Patronov, A.; Bjerrum, E. J.; Tyrchan, C.; Reymond, J.-L.; Chen, H.; Engkvist, O. SMILES-based deep generative scaffold decorator for de novo drug design. J. Cheminf. 2020, 12, 38,  DOI: 10.1186/s13321-020-00441-8
    [Crossref], [CAS], Google Scholar
    35
    SMILES-based deep generative scaffold decorator for de-novo drug design
    Arus-Pous, Josep; Patronov, Atanas; Bjerrum, Esben Jannik; Tyrchan, Christian; Reymond, Jean-Louis; Chen, Hongming; Engkvist, Ola
    Journal of Cheminformatics (2020), 12 (1), 38CODEN: JCOHB3; ISSN:1758-2946. (SpringerOpen)
    Herein we report a new SMILES-based mol. generative architecture that generates mols. from scaffolds and can be trained from any arbitrary mol. set. This approach is possible thanks to a new mol. set pre-processing algorithm that exhaustively slices all possible combinations of acyclic bonds of every mol., combinatorically obtaining a large no. of scaffolds with their resp. decorations. Two examples showcasing the potential of the architecture in medicinal and synthetic chem. are described: First, models were trained with a training set obtained from a small set of Dopamine Receptor D2 active modulators and were able to meaningfully decorate a wide range of scaffolds and obtain mol. series predicted active on DRD2. Second, a larger set of drug-like mols. from ChEMBL was selectively sliced using synthetic chem. constraints. This filtering process allowed models trained with this dataset to selectively decorate diverse scaffolds with fragments that were generally predicted to be synthesizable and attachable to the scaffold using known synthetic approaches. In both cases, the models were already able to decorate mols. using specific knowledge without the need to add it with other techniques, such as reinforcement learning. We envision that this architecture will become a useful addn. to the already existent architectures for de novo mol. generation.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVyksLjK&md5=05a686a14f942874fd6f6ddbc85017a1
  36. 36
    Fialková, V.; Zhao, J.; Papadopoulos, K.; Engkvist, O.; Bjerrum, E. J.; Kogej, T.; Patronov, A. LibINVENT: reaction-based generative scaffold decoration for in silico library design. J. Chem. Inf. Model. 2022, 62, 2046– 2063,  DOI: 10.1021/acs.jcim.1c00469
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    36
    LibINVENT: Reaction-based Generative Scaffold Decoration for in Silico Library Design
    Fialkova, Vendy; Zhao, Jiaxi; Papadopoulos, Kostas; Engkvist, Ola; Bjerrum, Esben Jannik; Kogej, Thierry; Patronov, Atanas
    Journal of Chemical Information and Modeling (2022), 62 (9), 2046-2063CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Because of the strong relationship between the desired mol. activity and its structural core, the screening of focused, core-sharing chem. libraries is a key step in lead optimization. Despite the plethora of current research focused on in silico methods for mol. generation, to our knowledge, no tool capable of designing such libraries has been proposed. In this work, we present a novel tool for de novo drug design called LibINVENT. It is capable of rapidly proposing chem. libraries of compds. sharing the same core while maximizing a range of desirable properties. To further help the process of designing focused libraries, the user can list specific chem. reactions that can be used for the library creation. LibINVENT is therefore a flexible tool for generating virtual chem. libraries for lead optimization in a broad range of scenarios. Addnl., the shared core ensures that the compds. in the library are similar, possess desirable properties, and can also be synthesized under the same or similar conditions. The LibINVENT code is freely available in our public repository at https://github.com/MolecularAI/Lib-INVENT. The code necessary for data preprocessing is further available at: https://github.com/MolecularAI/Lib-INVENT-dataset.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvV2iu7vI&md5=dc7b0da044f6aa1c2554a43e0aa14672
  37. 37
    Li, Y.; Hu, J.; Wang, Y.; Zhou, J.; Zhang, L.; Liu, Z. DeepScaffold: a comprehensive tool for scaffold-based de novo drug discovery using deep learning. J. Chem. Inf. Model. 2020, 60, 77– 91,  DOI: 10.1021/acs.jcim.9b00727
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    37
    DeepScaffold: A Comprehensive Tool for Scaffold-Based De Novo Drug Discovery Using Deep Learning
    Li, Yibo; Hu, Jianxing; Wang, Yanxing; Zhou, Jielong; Zhang, Liangren; Liu, Zhenming
    Journal of Chemical Information and Modeling (2020), 60 (1), 77-91CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    The ultimate goal of drug design is to find novel compds. with desirable pharmacol. properties. Designing mols. retaining particular scaffolds as their core structures is an efficient way to obtain potential drug candidates. We propose a scaffold-based mol. generative model for drug discovery, which performs mol. generation based on a wide spectrum of scaffold definitions, including Bemis-Murcko scaffolds, cyclic skeletons, and scaffolds with specifications on side-chain properties. The model can generalize the learned chem. rules of adding atoms and bonds to a given scaffold. The generated compds. were evaluated by mol. docking in DRD2 targets, and the results demonstrated that this approach can be effectively applied to solve several drug design problems, including the generation of compds. contg. a given scaffold and de novo drug design of potential drug candidates with specific docking scores.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXitlerurnN&md5=44ba7010e4824a37d97c66eed5bd1d6d
  38. 38
    Yang, Y.; Zheng, S.; Su, S.; Zhao, C.; Xu, J.; Chen, H. SyntaLinker: automatic fragment linking with deep conditional transformer neural networks. Chem. Sci. 2020, 11, 8312– 8322,  DOI: 10.1039/D0SC03126G
    [Crossref], [PubMed], [CAS], Google Scholar
    38
    SyntaLinker: automatic fragment linking with deep conditional transformer neural networks
    Yang, Yuyao; Zheng, Shuangjia; Su, Shimin; Zhao, Chao; Xu, Jun; Chen, Hongming
    Chemical Science (2020), 11 (31), 8312-8322CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
    Linking fragments to generate a focused compd. library for a specific drug target is one of the challenges in fragment-based drug design (FBDD). Hereby, we propose a new program named SyntaLinker, which is based on a syntactic pattern recognition approach using deep conditional transformer neural networks. This state-of-the-art transformer can link mol. fragments automatically by learning from the knowledge of structures in medicinal chem. databases (e.g.ChEMBL database). Conventionally, linking mol. fragments was viewed as connecting substructures that were predefined by empirical rules. In SyntaLinker, however, the rules of linking fragments can be learned implicitly from known chem. structures by recognizing syntactic patterns embedded in SMILES notations. With deep conditional transformer neural networks, SyntaLinker can generate mol. structures based on a given pair of fragments and addnl. restrictions. Case studies have demonstrated the advantages and usefulness of SyntaLinker in FBDD.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhsVagsrvK&md5=403cc7a09f96ecbe27afa3ce6bee51a2
  39. 39
    Feng, Y.; Yang, Y.; Deng, W.; Chen, H.; Ran, T. SyntaLinker-Hybrid: a deep learning approach for target specific drug design. Artif. Intell. Life Sci. 2022, 2, 100035,  DOI: 10.1016/j.ailsci.2022.100035
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  40. 40
    Huang, Y.; Peng, X.; Ma, J.; Zhang, M. 3DLinker: an E(3) equivariant variational autoencoder for molecular linker design. arXiv Preprint , arXiv:2205.07309, 2022.
    Google Scholar
    There is no corresponding record for this reference.
  41. 41
    Imrie, F.; Bradley, A. R.; van der Schaar, M.; Deane, C. M. Deep generative models for 3D linker design. J. Chem. Inf. Model. 2020, 60, 1983– 1995,  DOI: 10.1021/acs.jcim.9b01120
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    41
    Deep Generative Models for 3D Linker Design
    Imrie, Fergus; Bradley, Anthony R.; van der Schaar, Mihaela; Deane, Charlotte M.
    Journal of Chemical Information and Modeling (2020), 60 (4), 1983-1995CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    Rational compd. design remains a challenging problem for both computational methods and medicinal chemists. Computational generative methods have begun to show promising results for the design problem. However, they have not yet used the power of three-dimensional (3D) structural information. We have developed a novel graph-based deep generative model that combines state-of-the-art machine learning techniques with structural knowledge. Our method ("DeLinker") takes two fragments or partial structures and designs a mol. incorporating both. The generation process is protein-context-dependent, utilizing the relative distance and orientation between the partial structures. This 3D information is vital to successful compd. design, and we demonstrate its impact on the generation process and the limitations of omitting such information. In a large-scale evaluation, DeLinker designed 60% more mols. with high 3D similarity to the original mol. than a database baseline. When considering the more relevant problem of longer linkers with at least five atoms, the outperformance increased to 200%. We demonstrate the effectiveness and applicability of this approach on a diverse range of design problems: fragment linking, scaffold hopping, and proteolysis targeting chimera (PROTAC) design. As far as we are aware, this is the first mol. generative model to incorporate 3D structural information directly in the design process. The code is available at https://github.com/oxpig/DeLinker.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXltlenurs%253D&md5=430c929761c5020d1013538e028ba1bc
  42. 42
    Andrianov, G. V.; Gabriel Ong, W. J.; Serebriiskii, I.; Karanicolas, J. Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging. J. Chem. Inf. Model. 2021, 61, 5967– 5987,  DOI: 10.1021/acs.jcim.1c00630
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    42
    Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging
    Andrianov, Grigorii V.; Gabriel Ong, Wern Juin; Serebriiskii, Ilya; Karanicolas, John
    Journal of Chemical Information and Modeling (2021), 61 (12), 5967-5987CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    In early-stage drug discovery, the hit-to-lead optimization (or "hit expansion") stage entails starting from a newly identified active compd. and improving its potency or other properties. Traditionally, this process relies on synthesizing and evaluating a series of analogs to build up structure-activity relationships. Here, we describe a computational strategy focused on kinase inhibitors, intended to expedite the process of identifying analogs with improved potency. Our protocol begins from an inhibitor of the target kinase and generalizes the synthetic route used to access it. By searching for com. available replacements for the individual building blocks used to make the parent inhibitor, we compile an enumerated library of compds. that can be accessed using the same chem. transformations; these huge libraries can exceed many millions-or billions-of compds. Because the resulting libraries are much too large for explicit virtual screening, we instead consider alternate approaches to identify the top-scoring compds. We find that contributions from individual substituents are well described by a pairwise additivity approxn., provided that the corresponding fragments position their shared core in precisely the same way relative to the binding site. This key insight allows us to det. which fragments are suitable for merging into single new compds. and which are not. Further, the use of pairwise approxn. allows interaction energies to be assigned to each compd. in the library without the need for any further structure-based modeling: interaction energies instead can be reliably estd. from the energies of the component fragments, and the reduced computational requirements allow for flexible energy minimizations that allow the kinase to respond to each substitution. We demonstrate this protocol using libraries built from six representative kinase inhibitors drawn from the literature, which target five different kinases: CDK9, CHK1, CDK2, EGFRT790M, and ACK1. In each example, the enumerated library includes addnl. analogs reported by the original study to have activity, and these analogs are successfully prioritized within the library. We envision that the insights from this work can facilitate the rapid assembly and screening of increasingly large libraries for focused hit-to-lead optimization. To enable adoption of these methods and to encourage further analyses, we disseminate the computational tools needed to deploy this protocol.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisVGrs7zE&md5=d00fcbed94deb5955f19c3ee8d26af39
  43. 43
    Hall, R. J.; Murray, C. W.; Verdonk, M. L. The Fragment Network: a chemistry recommendation engine built using a graph database. J. Med. Chem. 2017, 60, 6440– 6450,  DOI: 10.1021/acs.jmedchem.7b00809
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    43
    The Fragment Network: A Chemistry Recommendation Engine Built Using a Graph Database
    Hall, Richard J.; Murray, Christopher W.; Verdonk, Marcel L.
    Journal of Medicinal Chemistry (2017), 60 (14), 6440-6450CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    The hit validation stage of a fragment-based drug discovery campaign involves probing the SAR around one or more fragment hits. This often requires a search for similar compds. in a corporate collection or from com. suppliers. The Fragment Network is a graph database that allows a user to efficiently search chem. space around a compd. of interest. The result set is chem. intuitive, naturally grouped by substitution pattern and meaningfully sorted according to the no. of observations of each transformation in medicinal chem. databases. This paper describes the algorithms used to construct and search the Fragment Network and provides examples of how it may be used in a drug discovery context.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXhtFKlsr3J&md5=79842771a068cdaabc2f9a841d791c1c
  44. 44
    Landrum, G. RDKit: A software suite for cheminformatics, computational chemistry, and predictive modeling. https://www.rdkit.org/ (accessed November 2022).
    Google Scholar
    There is no corresponding record for this reference.
  45. 45
    Neo4j, Graph Modeling Guidelines. https://neo4j.com/developer/guide-data-modeling/ (accessed November 2022).
    Google Scholar
    There is no corresponding record for this reference.
  46. 46
    Newman, J. A.; Douangamath, A.; Yadzani, S.; Yosaatmadja, Y.; Aimon, A.; Brandão-Neto, J.; Dunnett, L.; Gorrie-stone, T.; Skyner, R.; Fearon, D.; Schapira, M.; von Delft, F.; Gileadi, O. Structure, mechanism and crystallographic fragment screening of the SARS-CoV-2 NSP13 helicase. Nat. Commun. 2021, 12, 4848,  DOI: 10.1038/s41467-021-25166-6
    [Crossref], [PubMed], [CAS], Google Scholar
    46
    Structure, mechanism and crystallographic fragment screening of the SARS-CoV-2 NSP13 helicase
    Newman, Joseph A.; Douangamath, Alice; Yadzani, Setayesh; Yosaatmadja, Yuliana; Aimon, Antony; Brandao-Neto, Jose; Dunnett, Louise; Gorrie-stone, Tyler; Skyner, Rachael; Fearon, Daren; Schapira, Matthieu; von Delft, Frank; Gileadi, Opher
    Nature Communications (2021), 12 (1), 4848CODEN: NCAOBW; ISSN:2041-1723. (Nature Research)
    Abstr.: There is currently a lack of effective drugs to treat people infected with SARS-CoV-2, the cause of the global COVID-19 pandemic. The SARS-CoV-2 Non-structural protein 13 (NSP13) has been identified as a target for anti-virals due to its high sequence conservation and essential role in viral replication. Structural anal. reveals two "druggable" pockets on NSP13 that are among the most conserved sites in the entire SARS-CoV-2 proteome. Here we present crystal structures of SARS-CoV-2 NSP13 solved in the APO form and in the presence of both phosphate and a non-hydrolysable ATP analog. Comparisons of these structures reveal details of conformational changes that provide insights into the helicase mechanism and possible modes of inhibition. To identify starting points for drug development we have performed a crystallog. fragment screen against NSP13. The screen reveals 65 fragment hits across 52 datasets opening the way to structure guided development of novel antiviral agents.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhslOns7%252FI&md5=3e643800631b8bb56328c4ddd97fb7a7
  47. 47
    Jin, Z.; Du, X.; Xu, Y.; Deng, Y.; Liu, M.; Zhao, Y.; Zhang, B.; Li, X.; Zhang, L.; Peng, C.; Duan, Y.; Yu, J.; Wang, L.; Yang, K.; Liu, F.; Jiang, R.; Yang, X.; You, T.; Liu, X.; Yang, X.; Bai, F.; Liu, H.; Liu, X.; Guddat, L. W.; Xu, W.; Xiao, G.; Qin, C.; Shi, Z.; Jiang, H.; Rao, Z.; Yang, H. Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors. Nature 2020, 582, 289– 293,  DOI: 10.1038/s41586-020-2223-y
    [Crossref], [PubMed], [CAS], Google Scholar
    47
    Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors
    Jin, Zhenming; Du, Xiaoyu; Xu, Yechun; Deng, Yongqiang; Liu, Meiqin; Zhao, Yao; Zhang, Bing; Li, Xiaofeng; Zhang, Leike; Peng, Chao; Duan, Yinkai; Yu, Jing; Wang, Lin; Yang, Kailin; Liu, Fengjiang; Jiang, Rendi; Yang, Xinglou; You, Tian; Liu, Xiaoce; Yang, Xiuna; Bai, Fang; Liu, Hong; Liu, Xiang; Guddat, Luke W.; Xu, Wenqing; Xiao, Gengfu; Qin, Chengfeng; Shi, Zhengli; Jiang, Hualiang; Rao, Zihe; Yang, Haitao
    Nature (London, United Kingdom) (2020), 582 (7811), 289-293CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
    Abstr.: A new coronavirus, known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is the etiol. agent responsible for the 2019-2020 viral pneumonia outbreak of coronavirus disease 2019 (COVID-19). Currently, there are no targeted therapeutic agents for the treatment of this disease, and effective treatment options remain very limited. Here, we describe the results of a program that aimed to rapidly discover lead compds. for clin. use, by combining structure-assisted drug design, virtual drug screening and high-throughput screening. This program focused on identifying drug leads that target main protease (Mpro) of SARS-CoV-2: Mpro is a key enzyme of coronaviruses and has a pivotal role in mediating viral replication and transcription, making it an attractive drug target for SARS-CoV-2. We identified a mechanism-based inhibitor (N3) by computer-aided drug design, and then detd. the crystal structure of Mpro of SARS-CoV-2 in complex with this compd. Through a combination of structure-based virtual and high-throughput screening, we assayed more than 10,000 compds.-including approved drugs, drug candidates in clin. trials and other pharmacol. active compds.-as inhibitors of Mpro. Six of these compds. inhibited Mpro, showing half-maximal inhibitory concn. values that ranged from 0.67 to 21.4μM. One of these compds. (ebselen) also exhibited promising antiviral activity in cell-based assays. Our results demonstrate the efficacy of our screening strategy, which can lead to the rapid discovery of drug leads with clin. potential in response to new infectious diseases for which no specific drugs or vaccines are available.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVyhsrrO&md5=b84f350fe9ce1109485df6caf814ba82
  48. 48
    The COVID Moonshot Consortium. Achdout, H.; Aimon, A.; Bar-David, E.; Barr, H.; Ben-Shmuel, A.; Bennett, J.; Bilenko, V. A.; Bilenko, V. A.; Boby, M. L.; Borden, B.; Bowman, G. R.; Brun, J.; BVNBS, S.; Calmiano, M.; Carbery, A.; Carney, D.; Cattermole, E.; Chang, E.; Chernyshenko, E.; Chodera, J. D.; Clyde, A.; Coffland, J. E.; Cohen, G.; Cole, J.; Contini, A.; Cox, L.; Cvitkovic, M.; Dias, A.; Donckers, K.; Dotson, D. L.; Douangamath, A.; Duberstein, S.; Dudgeon, T.; Dunnett, L.; Eastman, P. K.; Erez, N.; Eyermann, C. J.; Fairhead, M.; Fate, G.; Fearon, D.; Fedorov, O.; Ferla, M.; Fernandes, R. S.; Ferrins, L.; Foster, R.; Foster, H.; Gabizon, R.; Garcia-Sastre, A.; Gawriljuk, V. O.; Gehrtz, P.; Gileadi, C.; Giroud, C.; Glass, W. G.; Glen, R.; Glinert, I.; Godoy, A. S.; Gorichko, M.; Gorrie-Stone, T.; Griffen, E. J.; Hart, S. H.; Heer, J.; Henry, M.; Hill, M.; Horrell, S.; Huliak, V. D.; Hurley, M. F.; Israely, T.; Jajack, A.; Jansen, J.; Jnoff, E.; Jochmans, D.; John, T.; Jonghe, S. D.; Kantsadi, A. L.; Kenny, P. W.; Kiappes, J. L.; Kinakh, S. O.; Koekemoer, L.; Kovar, B.; Krojer, T.; Lee, A.; Lefker, B. A.; Levy, H.; Logvinenko, I. G.; London, N.; Lukacik, P.; Macdonald, H. B.; MacLean, B.; Malla, T. R.; Matviiuk, T.; McCorkindale, W.; McGovern, B. L.; Melamed, S.; Melnykov, K. P.; Michurin, O.; Mikolajek, H.; Milne, B. F.; Morris, A.; Morris, G. M.; Morwitzer, M. J.; Moustakas, D.; Nakamura, A. M.; Neto, J. B.; Neyts, J.; Nguyen, L.; Noske, G. D.; Oleinikovas, V.; Oliva, G.; Overheul, G. J.; Owen, D.; Pai, R.; Pan, J.; Paran, N.; Perry, B.; Pingle, M.; Pinjari, J.; Politi, B.; Powell, A.; Psenak, V.; Puni, R.; Rangel, V. L.; Reddi, R. N.; Reid, S. P.; Resnick, E.; Ripka, E. G.; Robinson, M. C.; Robinson, R. P.; Rodriguez-Guerra, J.; Rosales, R.; Rufa, D.; Saar, K.; Saikatendu, K. S.; Schofield, C.; Shafeev, M.; Shaikh, A.; Shi, J.; Shurrush, K.; Singh, S.; Sittner, A.; Skyner, R.; Smalley, A.; Smeets, B.; Smilova, M. D.; Solmesky, L. J.; Spencer, J.; Strain-Damerell, C.; Swamy, V.; Tamir, H.; Tennant, R.; Thompson, W.; Thompson, A.; Tomasio, S.; Tsurupa, I. S.; Tumber, A.; Vakonakis, I.; van Rij, R. P.; Vangeel, L.; Varghese, F. S.; Vaschetto, M.; Vitner, E. B.; Voelz, V.; Volkamer, A.; von Delft, F.; von Delft, A.; Walsh, M.; Ward, W.; Weatherall, C.; Weiss, S.; White, K. M.; Wild, C. F.; Wittmann, M.; Wright, N.; Yahalom-Ronen, Y.; Zaidmann, D.; Zidane, H.; Zitzmann, N. Open science discovery of oral non-covalent SARS-CoV-2 main protease inhibitor therapeutics. bioRxiv Preprint , updated version, 2022.  DOI: 10.1101/2020.10.29.339317
    [Crossref], Google Scholar
    There is no corresponding record for this reference.
  49. 49
    Wahlberg, E.; Karlberg, T.; Kouznetsova, E.; Markova, N.; Macchiarulo, A.; Thorsell, A.-G.; Pol, E.; Frostell, s.; Ekblad, T.; Öncü, D.; Kull, B.; Robertson, G. M.; Pellicciari, R.; Schüler, H.; Weigelt, J. Family-wide chemical profiling and structural analysis of PARP and tankyrase inhibitors. Nat. Biotechnol. 2012, 30, 283– 288,  DOI: 10.1038/nbt.2121
    [Crossref], [PubMed], [CAS], Google Scholar
    49
    Family-wide chemical profiling and structural analysis of PARP and tankyrase inhibitors
    Wahlberg, Elisabet; Karlberg, Tobias; Kouznetsova, Ekaterina; Markova, Natalia; Macchiarulo, Antonio; Thorsell, Ann-Gerd; Pol, Ewa; Frostell, Aasa; Ekblad, Torun; Oencue, Delal; Kull, Bjoern; Robertson, Graeme Michael; Pellicciari, Roberto; Schueler, Herwig; Weigelt, Johan
    Nature Biotechnology (2012), 30 (3), 283-288CODEN: NABIF9; ISSN:1087-0156. (Nature Publishing Group)
    Inhibitors of poly-ADP-ribose polymerase (PARP) family proteins are currently in clin. trials as cancer therapeutics, yet the specificity of many of these compds. is unknown. Here we evaluated a series of 185 small-mol. inhibitors, including research reagents and compds. being tested clin., for the ability to bind to the catalytic domains of 13 of the 17 human PARP family members including the tankyrases, TNKS1 and TNKS2. Many of the best-known inhibitors, including TIQ-A, 6(5H)-phenanthridinone, olaparib, ABT-888 and rucaparib, bound to several PARP family members, suggesting that these mols. lack specificity and have promiscuous inhibitory activity. We also detd. X-ray crystal structures for five TNKS2 ligand complexes and four PARP14 ligand complexes. In addn. to showing that the majority of PARP inhibitors bind multiple targets, these results provide insight into the design of new inhibitors.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XisVKrsb4%253D&md5=71f17fd5b6bb602dd70c8fcc17df95f9
  50. 50
    Ohara-Nemoto, Y.; Shimoyama, Y.; Kimura, S.; Kon, A.; Haraga, H.; Ono, T.; Nemoto, T. K. Asp- and Glu-specific novel dipeptidyl peptidase 11 of Porphyromonas gingivalis ensures utilization of proteinaceous energy sources. J. Biol. Chem. 2011, 286, 38115– 38127,  DOI: 10.1074/jbc.M111.278572
    [Crossref], [PubMed], [CAS], Google Scholar
    50
    Asp- and Glu-specific Novel Dipeptidyl Peptidase 11 of Porphyromonas gingivalis Ensures Utilization of Proteinaceous Energy Sources
    Ohara-Nemoto, Yuko; Shimoyama, Yu; Kimura, Shigenobu; Kon, Asako; Haraga, Hiroshi; Ono, Toshio; Nemoto, Takayuki K.
    Journal of Biological Chemistry (2011), 286 (44), 38115-38127, S38115/1-S38115/9CODEN: JBCHA3; ISSN:0021-9258. (American Society for Biochemistry and Molecular Biology)
    Porphyromonas gingivalis and Porphyromonas endodontalis, asaccharolytic black-pigmented anaerobes, are predominant pathogens of human chronic and periapical periodontitis, resp. They incorporate di- and tripeptides from the environment as carbon and energy sources. In the present study we cloned a novel dipeptidyl peptidase (DPP) gene of P. endodontalis ATCC 35406, designated as DPP11. The DPP11 gene encoded 717 amino acids with a mol. mass of 81,090 Da and was present as a 75-kDa form with an N terminus of Asp22. A homol. search revealed the presence of a P. gingivalis ortholog, PGN0607, that has been categorized as an isoform of authentic DPP7. P. gingivalis DPP11 was exclusively cell-assocd. as a truncated 60-kDa form, and the gene ablation retarded cell growth. DPP11 specifically removed dipeptides from oligopeptides with the penultimate N-terminal Asp and Glu and has a P2-position preference to hydrophobic residues. Optimum pH was 7.0, and the kcat/Km value was higher for Asp than Glu. Those activities were lost by substitution of Ser652 in P. endodontalis and Ser655 in P. gingivalis DPP11 to Ala, and they were consistently decreased with increasing NaCl concn. Arg670 is a unique amino acid completely conserved in all DPP11 members distributed in the genera Porphyromonas, Bacteroides, and Parabacteroides, whereas this residue is converted to Gly in all authentic DPP7 members. Substitution anal. suggested that Arg670 interacts with an acidic residue of the substrate. Considered to preferentially utilize acidic amino acids, DPP11 ensures efficient degrdn. of oligopeptide substrates in these Gram-neg. anaerobic rods.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtlyktb%252FK&md5=419046327fe3bfdfb5619d722d9f9755
  51. 51
    Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Delivery Rev. 1997, 23, 3– 25,  DOI: 10.1016/S0169-409X(96)00423-1
    [Crossref], [CAS], Google Scholar
    51
    Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings
    Lipinski, Christopher A.; Lombardo, Franco; Dominy, Beryl W.; Feeney, Paul J.
    Advanced Drug Delivery Reviews (1997), 23 (1-3), 3-25CODEN: ADDREP; ISSN:0169-409X. (Elsevier)
    A review with 50 refs. Exptl. and computational approaches to est. soly. and permeability in discovery and development settings are described. In the discovery setting 'the rule of 5' predicts that poor absorption or permeation is more likely when there are more than 5 H-bond donors, 10 H-bond acceptors, the mol. wt. (MWT) is >500 and the calcd. Log P (CLogP) is >5. Computational methodol. for the rule-based Moriguchi Log P (MLogP) calcn. is described. Turbidimetric soly. measurement is described and applied to known drugs. High throughput screening (HTS) leads tend to have higher MWT and Log P and lower turbidimetric soly. than leads in the pre-HTS era. In the development setting, soly. calcns. focus on exact value prediction and are difficult because of polymorphism. Recent work on linear free energy relationships and Log P approaches are critically reviewed. Useful predictions are possible in closely related analog series when coupled with exptl. thermodn. soly. measurements.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2sXktlKlsQ%253D%253D&md5=405f70b0594d428f1275e1d56642cd3a
  52. 52
    Veber, D. F.; Johnson, S. R.; Cheng, H.-Y.; Smith, B. R.; Ward, K. W.; Kopple, K. D. Molecular properties that influence the oral bioavailability of drug candidates. J. Med. Chem. 2002, 45, 2615– 2623,  DOI: 10.1021/jm020017n
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    52
    Molecular Properties That Influence the Oral Bioavailability of Drug Candidates
    Veber, Daniel F.; Johnson, Stephen R.; Cheng, Hung-Yuan; Smith, Brian R.; Ward, Keith W.; Kopple, Kenneth D.
    Journal of Medicinal Chemistry (2002), 45 (12), 2615-2623CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Oral bioavailability measurements in rats for over 1100 drug candidates studied at Smith-Kline Beecham Pharmaceuticals (now Glaxo Smith-Kline) have allowed us to analyze the relative importance of mol. properties considered to influence that drug property. Reduced mol. flexibility, as measured by the no. of rotatable bonds, and low polar surface area or total hydrogen bond count (sum of donors and acceptors) are found to be important predictors of good oral bioavailability, independent of mol. wt. That on av. both the no. of rotatable bonds and polar surface area or hydrogen bond count tend to increase with mol. wt. may in part explain the success of the mol. wt. parameter in predicting oral bioavailability. The commonly applied mol. wt. cutoff at 500 does not itself significantly sep. compds. with poor oral bioavailability from those with acceptable values in this extensive data set. Our observations suggest that compds. which meet only the 2 criteria of (1) 10 or fewer rotatable bonds and (2) polar surface area ≤140 Å2 (or 12 or fewer H-bond donors and acceptors) will have a high probability of good oral bioavailability in the rat. Data sets for the artificial membrane permeation rate and for clearance in the rat were also examd. Reduced polar surface area correlates better with increased permeation rate than does lipophilicity (C log P), and increased rotatable bond count has a neg. effect on the permeation rate. A threshold permeation rate is a prerequisite of oral bioavailability. The rotatable bond count does not correlate with the data examd. here for the in vivo clearance rate in the rat.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38XjsFCmt7g%253D&md5=eaad26ed6a259de82ad65a8834fc397d
  53. 53
    Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 2012, 40, D1100– D1107,  DOI: 10.1093/nar/gkr777
    [Crossref], [PubMed], [CAS], Google Scholar
    53
    ChEMBL: a large-scale bioactivity database for drug discovery
    Gaulton, Anna; Bellis, Louisa J.; Bento, A. Patricia; Chambers, Jon; Davies, Mark; Hersey, Anne; Light, Yvonne; McGlinchey, Shaun; Michalovich, David; Al-Lazikani, Bissan; Overington, John P.
    Nucleic Acids Research (2012), 40 (D1), D1100-D1107CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
    ChEMBL is an Open Data database contg. binding, functional and ADMET information for a large no. of drug-like bioactive compds. These data are manually abstracted from the primary published literature on a regular basis, then further curated and standardized to maximize their quality and utility across a wide range of chem. biol. and drug-discovery research problems. Currently, the database contains 5.4 million bioactivity measurements for more than 1 million compds. and 5200 protein targets. Access is available through a web-based interface, data downloads and web services at: https://www.ebi.ac.uk/chembldb.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs12htbjN&md5=aedf7793e1ca54b6a4fa272ea3ef7d0e
  54. 54
    Ebejer, J.-P.; Morris, G. M.; Deane, C. M. Freely available conformer generation methods: how good are they?. J. Chem. Inf. Model. 2012, 52, 1146– 1158,  DOI: 10.1021/ci2004658
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    54
    Freely Available Conformer Generation Methods: How Good Are They?
    Ebejer, Jean-Paul; Morris, Garrett M.; Deane, Charlotte M.
    Journal of Chemical Information and Modeling (2012), 52 (5), 1146-1158CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    A review. Conformer generation has important implications in cheminformatics, particularly in computational drug discovery where the quality of conformer generation software may affect the outcome of a virtual screening exercise. We examine the performance of four freely available small mol. conformer generation tools (Balloon, Confab, Frog2, and RDKit) alongside a com. tool (MOE). The aim of this study is 3-fold: (i) to identify which tools most accurately reproduce exptl. detd. structures; (ii) to examine the diversity of the generated conformational set; and (iii) to benchmark the computational time expended. These aspects were tested using a set of 708 drug-like mols. assembled from the OMEGA validation set and the Astex Diverse Set. These mols. have varying physicochem. properties and at least one known X-ray crystal structure. We found that RDKit and Confab are statistically better than other methods at generating low rmsd conformers to the known structure. RDKit is particularly suited for less flexible mols. while Confab, with its systematic approach, is able to generate conformers which are geometrically closer to the exptl. detd. structure for mols. with a large no. of rotatable bonds (≥10). In our tests RDKit also resulted as the second fastest method after Frog2. In order to enhance the performance of RDKit, we developed a postprocessing algorithm to build a diverse and representative set of conformers which also contains a close conformer to the known structure. Our anal. indicates that, with postprocessing, RDKit is a valid free alternative to com., proprietary software.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38Xlt1Snu7s%253D&md5=2990e7dcfc83d8c0f7671250c51aeb97
  55. 55
    Su, M.; Yang, Q.; Du, Y.; Feng, G.; Liu, Z.; Li, Y.; Wang, R. Comparative assessment of scoring functions: the CASF-2016 update. J. Chem. Inf. Model. 2019, 59, 895– 913,  DOI: 10.1021/acs.jcim.8b00545
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    55
    Comparative Assessment of Scoring Functions: The CASF-2016 Update
    Su, Minyi; Yang, Qifan; Du, Yu; Feng, Guoqin; Liu, Zhihai; Li, Yan; Wang, Renxiao
    Journal of Chemical Information and Modeling (2019), 59 (2), 895-913CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
    In structure-based drug design, scoring functions are often employed to evaluate protein-ligand interactions. A variety of scoring functions have been developed so far, and thus, some objective benchmarks are desired for assessing their strength and weakness. The comparative assessment of scoring functions (CASF) benchmark developed by us provides an answer to this demand. CASF is designed as a "scoring benchmark", where the scoring process is decoupled from the docking process to depict the performance of scoring function more precisely. Here, we describe the latest update of this benchmark, i.e., CASF-2016. Each scoring function is still evaluated by four metrics, including "scoring power", "ranking power", "docking power", and "screening power". Nevertheless, the evaluation methods have been improved considerably in several aspects. A new test set is compiled, which consists of 285 protein-ligand complexes with high-quality crystal structures and reliable binding consts. A panel of 25 scoring functions are tested on CASF-2016 as a demonstration. Our results reveal that the performance of current scoring functions is more promising in terms of docking power than scoring, ranking, and screening power. Scoring power is somewhat correlated with ranking power, so are docking power and screening power. The results obtained on CASF-2016 may provide valuable guidance for the end users to make smart choices among available scoring functions. Moreover, CASF is created as an open-access benchmark so that other researchers can utilize it to test a wider range of scoring functions. The complete CASF-2016 benchmark will be released on the PDBbind-CN web server (http://www.pdbbind-cn.org/casf.asp/) once this article is published.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXitlWhtLjM&md5=573295cfd2298a2dee933027041c3d9c
  56. 56
    Ferla, M. Fragmenstein. https://github.com/matteoferla/Fragmenstein (accessed November 2022).
    Google Scholar
    There is no corresponding record for this reference.
  57. 57
    Chaudhury, S.; Lyskov, S.; Gray, J. J. PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta. Bioinformatics 2010, 26, 689– 691,  DOI: 10.1093/bioinformatics/btq007
    [Crossref], [PubMed], [CAS], Google Scholar
    57
    PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta
    Chaudhury, Sidhartha; Lyskov, Sergey; Gray, Jeffrey J.
    Bioinformatics (2010), 26 (5), 689-691CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)
    Summary: PyRosetta is a stand-alone Python-based implementation of the Rosetta mol. modeling package that allows users to write custom structure prediction and design algorithms using the major Rosetta sampling and scoring functions. PyRosetta contains Python bindings to libraries that define Rosetta functions including those for accessing and manipulating protein structure, calcg. energies and running Monte Carlo-based simulations. PyRosetta can be used in two ways: (i) interactively, using iPython and (ii) script-based, using Python scripting. Interactive mode contains a no. of help features and is ideal for beginners while script-mode is best suited for algorithm development. PyRosetta has similar computational performance to Rosetta, can be easily scaled up for cluster applications and has been implemented for algorithms demonstrating protein docking, protein folding, loop modeling and design. Availability: PyRosetta is a stand-alone package available at http://www.pyrosetta.org under the Rosetta license which is free for academic and non-profit users. A tutorial, user's manual and sample scripts demonstrating usage are also available on the web site.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXis1Wnt70%253D&md5=261b3b6b59e4987911556ea5314d9b19
  58. 58
    Adasme, M. F.; Linnemann, K. L.; Bolz, S. N.; Kaiser, F.; Salentin, S.; Haupt, V.; Schroeder, M. PLIP 2021: expanding the scope of the protein–ligand interaction profiler to DNA and RNA. Nucleic Acids Res. 2021, 49, W530– W534,  DOI: 10.1093/nar/gkab294
    [Crossref], [PubMed], [CAS], Google Scholar
    58
    PLIP 2021 expanding scope of protein-ligand interaction profiler to DNA and RNA
    Adasme, Melissa F.; Linnemann, Katja L.; Bolz, Sarah Naomi; Kaiser, Florian; Salentin, Sebastian; Haupt, V. Joachim; Schroeder, Michael
    Nucleic Acids Research (2021), 49 (W1), W530-W534CODEN: NARHAD; ISSN:1362-4962. (Oxford University Press)
    With the growth of protein structure data, the anal. of mol. interactions between ligands and their target mols. is gaining importance. PLIP, the protein-ligand interaction profiler, detects and visualises these interactions and provides data in formats suitable for further processing. PLIP has proven very successful in applications ranging from the characterization of docking expts. to the assessment of novel ligand-protein complexes. Besides ligand-protein interactions, interactions with DNA and RNA play a vital role in many applications, such as drugs targeting DNA or RNA-binding proteins. To date, over 7% of all 3D structures in the Protein Data Bank include DNA or RNA. Therefore, we extended PLIP to encompass these important mols. We demonstrate the power of this extension with examples of a cancer drug binding to a DNA target, and an RNA-protein complex central to a neurol. disease.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvV2isLbL&md5=8f56ced733405aaa8f5325e3fa4b862c
  59. 59
    Carbery, A.; Skyner, R.; von Delft, F.; Deane, C. M. Fragment libraries designed to be functionally diverse recover protein binding information more efficiently than standard structurally diverse libraries. J. Med. Chem. 2022, 65, 11404– 11413,  DOI: 10.1021/acs.jmedchem.2c01004
    [ACS Full Text ACS Full Text], [CAS], Google Scholar
    59
    Fragment Libraries Designed to Be Functionally Diverse Recover Protein Binding Information More Efficiently Than Standard Structurally Diverse Libraries
    Carbery, Anna; Skyner, Rachael; von Delft, Frank; Deane, Charlotte M.
    Journal of Medicinal Chemistry (2022), 65 (16), 11404-11413CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
    Current fragment-based drug design relies on the efficient exploration of chem. space by using structurally diverse libraries of small fragments. However, structurally dissimilar compds. can exploit the same interactions and thus be functionally similar. Using three-dimensional structures of many fragments bound to multiple targets, we examd. if a better strategy for selecting fragments for screening libraries exists. We show that structurally diverse fragments can be described as functionally redundant, often making the same interactions. Ranking fragments by the no. of novel interactions they made, we show that functionally diverse selections of fragments substantially increase the amt. of information recovered for unseen targets compared to the amts. recovered by other methods of selection. Using these results, we design small functionally efficient libraries that can give significantly more information about new protein targets than similarly sized structurally diverse libraries. By covering more functional space, we can generate more diverse sets of drug leads.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XitFWktLnO&md5=20c210891befe38eb6ec658cba2a01b8
  60. 60
    Maaten, L. v. d.; Hinton, G. Visualizing data using t-SNE. J. Mach. Learn. Res. 2008, 9, 2579– 2605
    Google Scholar
    There is no corresponding record for this reference.
  61. 61
    Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V.; Vanderplas, J.; Passos, A.; Cournapeau, D.; Brucher, M.; Perrot, M.; Duchesnay, d. Scikit-learn: machine learning in Python. J. Mach. Learn. Res. 2011, 12, 2825– 2830
    Google Scholar
    There is no corresponding record for this reference.
  62. 62
    PostEra, Mpro activity data. https://covid.postera.ai/covid/activity_data (accessed November 2022).
    Google Scholar
    There is no corresponding record for this reference.
  63. 63
    Butina, D. Unsupervised data base clustering based on Daylight’s fingerprint and Tanimoto similarity: a fast and automated way to cluster small and large data sets. JCICS 1999, 39, 747– 750,  DOI: 10.1021/ci9803381
    [ACS Full Text ACS Full Text], Google Scholar
    There is no corresponding record for this reference.

Cited By

This article has not yet been cited by other publications.

Download PDF

  • Abstract

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 1

    Figure 1. Pipeline for identifying fragment merges. Fragment hits from crystallographic fragment screens are used for finding fragment merges. All possible pairs of compounds are enumerated for merging (removing those with high similarity). Both the Fragment Network and similarity search are used to identify fragment merges. The Fragment Network enumerates all possible substructures of one of the fragments in the merge while the other fragment is regarded as the seed fragment. A series of optional hops are made away from the seed fragment (up to a maximum of two), after which an expansion is made by incorporating a substructure from the other fragment. The similarity search finds merges by calculating the Tversky (Tv) similarity against every compound in the database using the Morgan fingerprint (2048 bits and radius 2). The Tversky calculation uses α and β values of 0.7 and 0.3, respectively. All compounds with a mean similarity ≥0.4 are retained. The merges pass through a series of 2D and 3D filters, including pose generation with Fragmenstein, to result in scored poses.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 2

    Figure 2. The Fragment Network identifies pure merges. (a) A fragment-merging opportunity for the main protease (Mpro) data set. Interactions are predicted using the protein–ligand interaction profiler (PLIP). Hydrogen bonds and π-stacking interactions are shown by cyan and magenta dotted lines, respectively. The PanDDA density is provided for the purple fragment in the Supporting Information (owing to the unusual conformation). (b) The linker-like merge (pose generated using Fragmenstein) joins substructures from partially overlapping fragments by a “linker-like” region, maintaining the hydrogen bond with THR-45; a change in orientation of the thiazole ring (with respect to the thiophene ring in the fragment) enables an additional π-stacking interaction with HIS-41. (c) A fragment-linking opportunity for Mpro. (d) The proposed compound maintains a hydrogen bond with PHE-140 and makes an additional bond with SER-144. It is worth noting that the linker group proposed by the Fragment Network is present in thioacetazone, an oral antibiotic. (e) The fragments and merge in a and b in 2D. (f) The fragments and merge in c and d in 2D.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 3

    Figure 3. The Fragment Network and similarity searches identify filtered compounds for different fragment pairs. The numbers of filtered compounds for each fragment pair found using the Fragment Network (blue) or similarity search (orange) are shown across targets (a) dipeptidyl peptidase 11 (DPP11), (b) poly(ADP-ribose) polymerase 14, (PARP14), (c) nonstructural protein 13 (nsp13), and (d) main protease (Mpro). Only pairs that resulted in filtered compounds are shown. Pairs are ordered from right to left according to the number of Fragment Network compounds found. The data show that each search technique was able to identify filtered compounds for pairs where the other technique identified none.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 4

    Figure 4. Fragment Network and similarity search-derived compound sets populate different regions of chemical space. The chemical space occupied by the filtered compound sets is projected into two dimensions using the T-SNE algorithm across targets (a) dipeptidyl peptidase 11 (DPP11), (b) poly(ADP-ribose) polymerase 14 (PARP14), (c) nonstructural protein 13 (nsp13), and (d) main protease (Mpro). Fragment Network compounds are shown in blue, and similarity search compounds are shown in orange. The two compound sets are shown to occupy distinct areas of chemical space.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 5

    Figure 5. The Fragment Network identifies a known binder against Mpro. A Fragment Network search using (a) two fragment hits against the SARS-CoV-2 main protease (Mpro) identifies, (b) a known binder against Mpro (LON-WEI-b2874fec-25; RapidFire mass spectrometry (RF-MS) IC50 value of 59.6 μM) and similar compounds to known binders, (c) JAN-GHE-83b26c96–22 (fluorescence and RF-MS IC50 values of 96.9 μM and 24.5 μM), and (d) TRY-UNI-714a760b-18 (fluorescence and RF-MS IC50 values of 26.2 μM and 13.0 μM). Fragmenstein-predicted merge poses are shown in white, and crystal poses are in cyan. Interactions are predicted using the protein–ligand interaction profiler (PLIP), and key interaction residues are shown. Hydrogen bonds are shown in cyan, and π-stacking interactions are shown in magenta.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 6

    Figure 6. The Fragment Network identifies a known binder against EthR. A Fragment Network search using (a) two fragment hits against (which each bind in two different positions) Mycobacterium tuberculosis transcriptional repressor protein EthR identifies (b) a known binder (compound 4; IC50 value of >100 μM). (c) An alternative crystallographic arrangement of the equivalent fragments identifies (d) a similar compound to a known binder, compound 21 (IC50 value of 22 μM). Fragmenstein-predicted merge poses are shown in white, and crystal poses are in cyan. Hydrogen bonds are shown in cyan.

    High Resolution Image
    Download MS PowerPoint Slide

  • References

    ARTICLE SECTIONS
    Jump To

    This article references 63 other publications.

    1. 1
      Lamoree, B.; Hubbard, R. Current perspectives in fragment-based lead discovery (FBLD). Essays Biochem 2017, 61, 453– 464,  DOI: 10.1042/EBC20170028
      [Crossref], [PubMed], [CAS], Google Scholar
      1
      Current perspectives in fragment-based lead discovery (FBLD)
      Lamoree Bas; Hubbard Roderick E; Hubbard Roderick E
      Essays in biochemistry (2017), 61 (5), 453-464 ISSN:.
      It is over 20 years since the first fragment-based discovery projects were disclosed. The methods are now mature for most 'conventional' targets in drug discovery such as enzymes (kinases and proteases) but there has also been growing success on more challenging targets, such as disruption of protein-protein interactions. The main application is to identify tractable chemical startpoints that non-covalently modulate the activity of a biological molecule. In this essay, we overview current practice in the methods and discuss how they have had an impact in lead discovery - generating a large number of fragment-derived compounds that are in clinical trials and two medicines treating patients. In addition, we discuss some of the more recent applications of the methods in chemical biology - providing chemical tools to investigate biological molecules, mechanisms and systems.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BC1M7pt1Gmtw%253D%253D&md5=35bce3c5978329c6737ffd719135d363
    2. 2
      Davis, B. J.; Roughley, S. D. In Platform Technologies in Drug Discovery and Validation; Goodnow, R. A., Ed.; Annual Reports in Medicinal Chemistry series; Academic Press, 2017; Vol. 50; pp 371– 439.
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    3. 3
      Hann, M. M.; Leach, A. R.; Harper, G. Molecular complexity and its impact on the probability of finding leads for drug discovery. J. Chem. Inf. Model. 2001, 41, 856– 864,  DOI: 10.1021/ci000403i
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      3
      Molecular Complexity and Its Impact on the Probability of Finding Leads for Drug Discovery
      Hann, Michael M.; Leach, Andrew R.; Harper, Gavin
      Journal of Chemical Information and Computer Sciences (2001), 41 (3), 856-864CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
      Using a simple model of ligand-receptor interactions, the interactions between ligands and receptors of varying complexities are studied and the probabilities of binding calcd. It is obsd. that as the systems become more complex the chance of observing a useful interaction for a randomly chosen ligand falls dramatically. The implications of this for the design of combinatorial libraries is explored. A large set of drug leads and optimized compds. is profiled using several different properties relevant to mol. recognition. The changes obsd. for these properties during the drug optimization phase support the hypothesis that less complex mols. are more common starting points for the discovery of drugs. An extreme example of the use of simple mols. for directed screening against thrombin is provided.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXivFynt7w%253D&md5=e89e76c8fb83fc5012369ae30f20e6b7
    4. 4
      Leach, A. R.; Hann, M. M. Molecular complexity and fragment-based drug discovery: ten years on. Curr. Opin. Chem. Biol. 2011, 15, 489– 496,  DOI: 10.1016/j.cbpa.2011.05.008
      [Crossref], [PubMed], [CAS], Google Scholar
      4
      Molecular complexity and fragment-based drug discovery: ten years on
      Leach, Andrew R.; Hann, Michael M.
      Current Opinion in Chemical Biology (2011), 15 (4), 489-496CODEN: COCBF4; ISSN:1367-5931. (Elsevier B.V.)
      We review the concept of mol. complexity in the context of the very simple model of mol. interactions that we introduced over 10 years ago. A summary is presented of efforts to validate this simple model using screening data. The relationship between the complexity model and the problem of sampling chem. space is discussed, together with the relevance of these theor. concepts to fragment-based drug discovery.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXpvFCltrg%253D&md5=19c2d65de80a8bab1059468828da6ce2
    5. 5
      Keserű, G. M.; Erlanson, D. A.; Ferenczy, G. G.; Hann, M. M.; Murray, C. W.; Pickett, S. D. Design principles for fragment libraries: maximizing the value of learnings from pharma fragment-based drug discovery (FBDD) programs for use in academia. J. Med. Chem. 2016, 59, 8189– 8206,  DOI: 10.1021/acs.jmedchem.6b00197
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      5
      Design Principles for Fragment Libraries: Maximizing the Value of Learnings from Pharma Fragment-Based Drug Discovery (FBDD) Programs for Use in Academia
      Keseru, Gyorgy M.; Erlanson, Daniel A.; Ferenczy, Gyorgy G.; Hann, Michael M.; Murray, Christopher W.; Pickett, Stephen D.
      Journal of Medicinal Chemistry (2016), 59 (18), 8189-8206CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Fragment-based drug discovery (FBDD) is well suited for discovering both drug leads and chem. probes of protein function; it can cover broad swaths of chem. space and allows the use of creative chem. FBDD is widely implemented for lead discovery in industry but is sometimes used less systematically in academia. Design principles and implementation approaches for fragment libraries are continually evolving, and the lack of up-to-date guidance may prevent more effective application of FBDD in academia. This Perspective explores many of the theor., practical, and strategic considerations that occur within FBDD programs, including the optimal size, complexity, physicochem. profile, and shape profile of fragments in FBDD libraries, as well as compd. storage, evaluation, and screening technologies. This compilation of industry experience in FBDD will hopefully be useful for those pursuing FBDD in academia.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XmvFSit7s%253D&md5=888fe2cf003c5564c2ec391cf54ea2ac
    6. 6
      Hall, R. J.; Mortenson, P. N.; Murray, C. W. Efficient exploration of chemical space by fragment-based screening. Prog. Biophys. Mol. Biol. 2014, 116, 82– 91,  DOI: 10.1016/j.pbiomolbio.2014.09.007
      [Crossref], [PubMed], [CAS], Google Scholar
      6
      Efficient exploration of chemical space by fragment-based screening
      Hall, Richard J.; Mortenson, Paul N.; Murray, Christopher W.
      Progress in Biophysics & Molecular Biology (2014), 116 (2-3), 82-91CODEN: PBIMAC; ISSN:0079-6107. (Elsevier Ltd.)
      Screening methods seek to sample a vast chem. space in order to identify starting points for further chem. optimization. Fragment based drug discovery exploits the superior sampling of chem. space that can be achieved when the mol. wt. is restricted. Here we show that com. available fragment space is still relatively poorly sampled and argue for highly sensitive screening methods to allow the detection of smaller fragments. We analyze the properties of our fragment library vs. the properties of X-ray hits derived from the library. We particularly consider properties related to the degree of planarity of the fragments.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1ymu7zO&md5=a63e451443207393b781d3d68b10bcd1
    7. 7
      Murray, C. W.; Rees, D. C. The rise of fragment-based drug discovery. Nat. Chem. 2009, 1, 187– 192,  DOI: 10.1038/nchem.217
      [Crossref], [PubMed], [CAS], Google Scholar
      7
      The rise of fragment-based drug discovery
      Murray, Christopher W.; Rees, David C.
      Nature Chemistry (2009), 1 (3), 187-192CODEN: NCAHBB; ISSN:1755-4330. (Nature Publishing Group)
      A review. The search for new drugs is plagued by high attrition rates at all stages in research and development. Chemists have an opportunity to tackle this problem because attrition can be traced back, in part, to the quality of the chem. leads. Fragment-based drug discovery (FBDD) is a new approach, increasingly used in the pharmaceutical industry, for reducing attrition and providing leads for previously intractable biol. targets. FBDD identifies low-mol.-wt. ligands (∼150 Da) that bind to biol. important macromols. The three-dimensional exptl. binding mode of these fragments is detd. using X-ray crystallog. or NMR spectroscopy, and is used to facilitate their optimization into potent mols. with drug-like properties. Compared with high-throughput-screening, the fragment approach requires fewer compds. to be screened, and, despite the lower initial potency of the screening hits, offers more efficient and fruitful optimization campaigns. Here, we review the rise of FBDD, including its application to discovering clin. candidates against targets for which other chem. approaches have struggled.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXmtF2rs78%253D&md5=0bd1a40fd927cce9200f0fa296a7462f
    8. 8
      Roughley, S. D.; Hubbard, R. E. How well can fragments explore accessed chemical space? A case study from heat shock protein 90. J. Med. Chem. 2011, 54, 3989– 4005,  DOI: 10.1021/jm200350g
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      8
      How Well Can Fragments Explore Accessed Chemical Space? A Case Study from Heat Shock Protein 90
      Roughley, Stephen D.; Hubbard, Roderick E.
      Journal of Medicinal Chemistry (2011), 54 (12), 3989-4005CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      A review. This article summerizes the published literature on the generation of inhibitors and design of drug against the mol. chaperone, heat shock protein 90 (Hsp90), as potential therapeutics to treat cancer.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXmsFymurs%253D&md5=90c2c8028cb460c0c6fdc956e7daae28
    9. 9
      Collins, P. M.; Douangamath, A.; Talon, R.; Dias, A.; Brandao-Neto, J.; Krojer, T.; von Delft, F. Achieving a good crystal system for crystallographic X-ray fragment screening. Methods Enzymol 2018, 610, 251– 264,  DOI: 10.1016/bs.mie.2018.09.027
      [Crossref], [PubMed], [CAS], Google Scholar
      9
      Achieving a good crystal system for crystallographic X-ray fragment screening
      Collins, Patrick M.; Douangamath, Alice; Talon, Romain; Dias, Alexandre; Brandao-Neto, Jose; Krojer, Tobias; von Delft, Frank
      Methods in Enzymology (2018), 610 (Modern Approaches in Drug Discovery), 251-264CODEN: MENZAU; ISSN:0076-6879. (Elsevier Inc.)
      A review. The XChem facility at Diamond Light Source offers fragment screening by X-ray crystallog. as a general access user program. The main advantage of X-ray crystallog. as a primary fragment screen is that it yields directly the location and pose of the fragment hits, whether within pockets of interest or merely on surface sites: this is the key information for structure-based design and for enabling synthesis of follow up mols. Extensive streamlining of the screening expt. at XChem has engendered a very active user program that is generating large amts. of data: in 2017,36 academic and industry groups generated 35,000 datasets of uniquely soaked crystals. It has also generated a large no. of learnings concerning the main remaining bottleneck namely, obtaining a suitable crystal system that will support a successful fragment screen. Here we discuss the practicalities of generating screen-ready crystals that have useful electron d. maps, and how to ensure they will be successfully reproduced and usable at a facility outside the home lab.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhsFKrsrvJ&md5=c122d61fb2371eff58429ce90e777c8e
    10. 10
      The COVID Moonshot Consortium. COVID Moonshot: open science discovery of SARS-CoV-2 main protease inhibitors by combining crowdsourcing, high-throughput experiments, computational simulations, and machine learning. bioRxiv Preprint , 2020.  DOI: 10.1101/2020.10.29.339317
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    11. 11
      Douangamath, A.; Powell, A.; Fearon, D.; Collins, P. M.; Talon, R.; Krojer, T.; Skyner, R.; Brandao-Neto, J.; Dunnett, L.; Dias, A.; Aimon, A.; Pearce, N. M.; Wild, C.; Gorrie-Stone, T.; von Delft, F. Achieving efficient fragment screening at XChem facility at Diamond Light Source. JoVE 2021, 171, e62414  DOI: 10.3791/62414
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    12. 12
      Metz, A.; Wollenhaupt, J.; Glöckner, S.; Messini, N.; Huber, S.; Barthel, T.; Merabet, A.; Gerber, H.-D.; Heine, A.; Klebe, G.; Weiss, M. S. Frag4Lead: growing crystallographic fragment hits by catalog using fragment-guided template docking. Acta Crystallogr., Sect. D: Struct. Biol. 2021, 77, 1168– 1182,  DOI: 10.1107/S2059798321008196
      [Crossref], [CAS], Google Scholar
      12
      Frag4Lead: growing crystallographic fragment hits by catalog using fragment-guided template docking
      Metz, Alexander; Wollenhaupt, Jan; Gloeckner, Steffen; Messini, Niki; Huber, Simon; Barthel, Tatjana; Merabet, Ahmed; Gerber, Hans-Dieter; Heine, Andreas; Klebe, Gerhard; Weiss, Manfred S.
      Acta Crystallographica, Section D: Structural Biology (2021), 77 (9), 1168-1182CODEN: ACSDAD; ISSN:2059-7983. (International Union of Crystallography)
      In recent years, crystallog. fragment screening has matured into an almost routine expt. at several modern synchrotron sites. The hits of the screening expt., i.e. small mols. or fragments binding to the target protein, are revealed along with their 3D structural information. Therefore, they can serve as useful starting points for further structure-based hit-to-lead development. However, the progression of fragment hits to tool compds. or even leads is often hampered by a lack of chem. feasibility. As an attractive alternative, compd. analogs that embed the fragment hit structurally may be obtained from com. catalogs. Here, a workflow is reported based on filtering and assessing such potential follow-up compds. by template docking. This means that the crystallog. binding pose was integrated into the docking calcns. as a central starting parameter. Subsequently, the candidates are scored on their interactions within the binding pocket. In an initial proof-of-concept study using five starting fragments known to bind to the aspartic protease endothiapepsin, 28 follow-up compds. were selected using the designed workflow and their binding was assessed by crystallog. Ten of these compds. bound to the active site and five of them showed significantly increased affinity in isothermal titrn. calorimetry of up to single-digit micromolar affinity. Taken together, this strategy is capable of efficiently evolving the initial fragment hits without major synthesis efforts and with full control by X-ray crystallog.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvFegt7%252FF&md5=391b172987e46364a4b877d2a34508fc
    13. 13
      Müller, J.; Klein, R.; Tarkhanova, O.; Gryniukova, A.; Borysko, P.; Merkl, S.; Ruf, M.; Neumann, A.; Gastreich, M.; Moroz, Y. S.; Klebe, G.; Glinca, S. Magnet for the needle in haystack: “Crystal structure first” Fragment hits unlock active chemical matter using targeted exploration of vast chemical spaces. J. Med. Chem. 2022, 65, 15663– 15678,  DOI: 10.1021/acs.jmedchem.2c00813
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      13
      Magnet for the Needle in Haystack: "Crystal Structure First" Fragment Hits Unlock Active Chemical Matter Using Targeted Exploration of Vast Chemical Spaces
      Mueller, Janis; Klein, Raphael; Tarkhanova, Olga; Gryniukova, Anastasiia; Borysko, Petro; Merkl, Stefan; Ruf, Moritz; Neumann, Alexander; Gastreich, Marcus; Moroz, Yurii S.; Klebe, Gerhard; Glinca, Serghei
      Journal of Medicinal Chemistry (2022), 65 (23), 15663-15678CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Fragment-based drug discovery (FBDD) has successfully led to approved therapeutics for challenging and "undruggable" targets. In the context of FBDD, we introduce a novel, multidisciplinary method to identify active mols. from purchasable chem. space. Starting from four small-mol. fragment complexes of protein kinase A (PKA), a template-based docking screen using Enamine's multibillion REAL Space was performed. A total of 93 mols. out of 106 selected compds. were successfully synthesized. Forty compds. were active in at least one validation assay with the most active follow-up having a 13,500-fold gain in affinity. Crystal structures for six of the most promising binders were rapidly obtained, verifying the binding mode. The overall success rate for this novel fragment-to-hit approach was 40%, accomplished in only 9 wk. The results challenge the established fragment prescreening paradigm since the std. industrial filters for fragment hit identification in a thermal shift assay would have missed the initial fragments.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XitlWrtrfI&md5=7d2ae978c7757c8f1b3083fa84c60f9e
    14. 14
      Piticchio, S. G.; Martínez-Cartró, M.; Scaffidi, S.; Rachman, M.; Rodriguez-Arevalo, S.; Sanchez-Arfelis, A.; Escolano, C.; Picaud, S.; Krojer, T.; Filippakopoulos, P.; von Delft, F.; Galdeano, C.; Barril, X. Discovery of novel BRD4 ligand scaffolds by automated navigation of the fragment chemical space. J. Med. Chem. 2021, 64, 17887– 17900,  DOI: 10.1021/acs.jmedchem.1c01108
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      14
      Discovery of Novel BRD4 Ligand Scaffolds by Automated Navigation of the Fragment Chemical Space
      Piticchio, Serena G.; Martinez-Cartro, Miriam; Scaffidi, Salvatore; Rachman, Moira; Rodriguez-Arevalo, Sergio; Sanchez-Arfelis, Ainoa; Escolano, Carmen; Picaud, Sarah; Krojer, Tobias; Filippakopoulos, Panagis; von Delft, Frank; Galdeano, Carles; Barril, Xavier
      Journal of Medicinal Chemistry (2021), 64 (24), 17887-17900CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Fragment-based drug discovery (FBDD) is a very effective hit identification method. However, the evolution of fragment hits into suitable leads remains challenging and largely artisanal. Fragment evolution is often scaffold-centric, meaning that its outcome depends crucially on the chem. structure of the starting fragment. Considering that fragment screening libraries cover only a small proportion of the corresponding chem. space, hits should be seen as probes highlighting privileged areas of the chem. space rather than actual starting points. We have developed an automated computational pipeline to mine the chem. space around any specific fragment hit, rapidly finding analogs that share a common interaction motif but are structurally novel and diverse. On a prospective application on the bromodomain-contg. protein 4 (BRD4), starting from a known fragment, the platform yields active mols. with nonobvious scaffold changes. The procedure is fast and inexpensive and has the potential to uncover many hidden opportunities in FBDD.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXislWrtbjE&md5=98aebd322c14abb1a7f56157e46013a6
    15. 15
      Miyake, Y.; Itoh, Y.; Hatanaka, A.; Suzuma, Y.; Suzuki, M.; Kodama, H.; Arai, Y.; Suzuki, T. Identification of novel lysine demethylase 5-selective inhibitors by inhibitor-based fragment merging strategy. Bioorg. Med. Chem. 2019, 27, 1119– 1129,  DOI: 10.1016/j.bmc.2019.02.006
      [Crossref], [PubMed], [CAS], Google Scholar
      15
      Identification of novel lysine demethylase 5-selective inhibitors by inhibitor-based fragment merging strategy
      Miyake, Yuka; Itoh, Yukihiro; Hatanaka, Atsushi; Suzuma, Yoshinori; Suzuki, Miki; Kodama, Hidehiko; Arai, Yoshinobu; Suzuki, Takayoshi
      Bioorganic & Medicinal Chemistry (2019), 27 (6), 1119-1129CODEN: BMECEP; ISSN:0968-0896. (Elsevier B.V.)
      Histone lysine demethylases (KDMs) have drawn much attention as targets of therapeutic agents. KDM5 proteins, which are Fe(II)/α-ketoglutarate-dependent demethylases, are assocd. with oncogenesis and drug resistance in cancer cells, and KDM5-selective inhibitors are expected to be anticancer drugs. However, few cell-active KDM5 inhibitors have been reported and there is an obvious need to discover more. In this study, we pursued the identification of highly potent and cell-active KDM5-selective inhibitors. Based on the reported KDM5 inhibitors, we designed several compds. by strategically merging two fragments for competitive inhibition with α-ketoglutarate and for KDM5-selective inhibition. Among them, compds. 10 and 13, which have a 3-cyano pyrazolo[1,5-a]pyrimidin-7-one scaffold, exhibited strong KDM5-inhibitory activity and significant KDM5 selectivity. In cellular assays using human lung cancer cell line A549, 10 and 13 increased the levels of trimethylated lysine 4 on histone H3, which is a specific substrate of KDM5s, and induced growth inhibition of A549 cells. These results should provide a basis for the development of cell-active KDM5 inhibitors to highlight the validity of our inhibitor-based fragment merging strategy.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXivVWltbg%253D&md5=1d9473e11de97ee34b82b82b8d6ffb38
    16. 16
      Ren, J.; Li, J.; Wang, Y.; Chen, W.; Shen, A.; Liu, H.; Chen, D.; Cao, D.; Li, Y.; Zhang, N.; Xu, Y.; Geng, M.; He, J.; Xiong, B.; Shen, J. Identification of a new series of potent diphenol HSP90 inhibitors by fragment merging and structure-based optimization. Bioorg. Med. Chem. Lett. 2014, 24, 2525– 2529,  DOI: 10.1016/j.bmcl.2014.03.100
      [Crossref], [PubMed], [CAS], Google Scholar
      16
      Identification of a new series of potent diphenol HSP90 inhibitors by fragment merging and structure-based optimization
      Ren, Jing; Li, Jian; Wang, Yueqin; Chen, Wuyan; Shen, Aijun; Liu, Hongchun; Chen, Danqi; Cao, Danyan; Li, Yanlian; Zhang, Naixia; Xu, Yechun; Geng, Meiyu; He, Jianhua; Xiong, Bing; Shen, Jingkang
      Bioorganic & Medicinal Chemistry Letters (2014), 24 (11), 2525-2529CODEN: BMCLE8; ISSN:0960-894X. (Elsevier B.V.)
      Heat shock protein 90 (HSP90) is a mol. chaperone to fold and maintain the proper conformation of many signaling proteins, esp. some oncogenic proteins and mutated unstable proteins. Inhibition of HSP90 was recognized as an effective approach to simultaneously suppress several aberrant signaling pathways, and therefore it was considered as a novel target for cancer therapy. Here, by integrating several techniques including the fragment-based drug discovery method, fragment merging, computer aided inhibitor optimization, and structure-based drug design, the authors were able to identify a series of HSP90 inhibitors, e.g. I [R1 = H. HO2CCH2, NH2CH2CH2, MeCONHCH2CH2; r2 = H, Br, O2N, etc.; R3 = H, Br, Me, MeO]. Several compds. can inhibit HSP90 with IC50 about 20-40 nM, which is at least 200-fold more potent than initial fragments in the protein binding assay. These new HSP90 inhibitors not only explore interactions with an under-studied subpocket, also offer new chemotypes for the development of novel HSP90 inhibitors as anticancer drugs.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXmslahsrY%253D&md5=057e15a1982cf723de8c7436accf4fff
    17. 17
      Credille, C. V.; Chen, Y.; Cohen, S. M. Fragment-based identification of influenza endonuclease inhibitors. J. Med. Chem. 2016, 59, 6444– 6454,  DOI: 10.1021/acs.jmedchem.6b00628
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      17
      Fragment-Based Identification of Influenza Endonuclease Inhibitors
      Credille, Cy V.; Chen, Yao; Cohen, Seth M.
      Journal of Medicinal Chemistry (2016), 59 (13), 6444-6454CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      The influenza virus is responsible for millions of cases of severe illness annually. Yearly variance in the effectiveness of vaccination, coupled with emerging drug resistance, necessitates the development of new drugs to treat influenza infections. One attractive target is the RNA-dependent RNA polymerase PA subunit. Herein we report the development of inhibitors of influenza PA endonuclease derived from lead compds. identified from a metal-binding pharmacophore (MBP) library screen. Pyromeconic acid and derivs. thereof were found to be potent inhibitors of endonuclease. Guided by modeling and previously reported structural data, several sublibraries of mols. were elaborated from the MBP hits. Structure-activity relationships were established, and more potent mols. were designed and synthesized using fragment growth and fragment merging strategies. This approach ultimately resulted in the development of a lead compd. with an IC50 value of 14 nM, which displayed an EC50 value of 2.1 μM against H1N1 influenza virus in MDCK cells.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XpsF2mtL8%253D&md5=a872083d373a846467c7f2eb5439bed4
    18. 18
      Edink, E.; Rucktooa, P.; Retra, K.; Akdemir, A.; Nahar, T.; Zuiderveld, O.; van Elk, R.; Janssen, E.; van Nierop, P.; van Muijlwijk-Koezen, J.; Smit, A. B.; Sixma, T. K.; Leurs, R.; de Esch, I. J. P. Fragment growing induces conformational changes in acetylcholine-binding protein: a structural and thermodynamic analysis. J. Am. Chem. Soc. 2011, 133, 5363– 5371,  DOI: 10.1021/ja110571r
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      18
      Fragment Growing Induces Conformational Changes in Acetylcholine-Binding Protein: A Structural and Thermodynamic Analysis
      Edink, Ewald; Rucktooa, Prakash; Retra, Kim; Akdemir, Atilla; Nahar, Tariq; Zuiderveld, Obbe; van Elk, Rene; Janssen, Elwin; van Nierop, Pim; van Muijlwijk-Koezen, Jacqueline; Smit, August B.; Sixma, Titia K.; Leurs, Rob; de Esch, Iwan J. P.
      Journal of the American Chemical Society (2011), 133 (14), 5363-5371CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
      Optimization of fragment hits toward high-affinity lead compds. is a crucial aspect of fragment-based drug discovery (FBDD). In the current study, we have successfully optimized a fragment by growing into a ligand-inducible subpocket of the binding site of acetylcholine-binding protein (AChBP). This protein is a sol. homolog of the ligand binding domain (LBD) of Cys-loop receptors. The fragment optimization was monitored with X-ray structures of ligand complexes and systematic thermodn. analyses using surface plasmon resonance (SPR) biosensor anal. and isothermal titrn. calorimetry (ITC). Using site-directed mutagenesis and AChBP from different species, we find that specific changes in thermodn. binding profiles, are indicative of interactions with the ligand-inducible subpocket of AChBP. This study illustrates that thermodn. anal. provides valuable information on ligand binding modes and is complementary to affinity data when guiding rational structure- and fragment-based discovery approaches.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhvFeisrc%253D&md5=413798f2128e309545de77f8e995b565
    19. 19
      Hughes, S. J.; Millan, D. S.; Kilty, I. C.; Lewthwaite, R. A.; Mathias, J. P.; O’Reilly, M. A.; Pannifer, A.; Phelan, A.; Stühmeier, F.; Baldock, D. A.; Brown, D. G. Fragment based discovery of a novel and selective PI3 kinase inhibitor. Bioorg. Med. Chem. Lett. 2011, 21, 6586– 6590,  DOI: 10.1016/j.bmcl.2011.07.117
      [Crossref], [PubMed], [CAS], Google Scholar
      19
      Fragment based discovery of a novel and selective PI3 kinase inhibitor
      Hughes, Samantha J.; Millan, David S.; Kilty, Iain C.; Lewthwaite, Russell A.; Mathias, John P.; O'Reilly, Mark A.; Pannifer, Andrew; Phelan, Anne; Stuehmeier, Frank; Baldock, Darren A.; Brown, David G.
      Bioorganic & Medicinal Chemistry Letters (2011), 21 (21), 6586-6590CODEN: BMCLE8; ISSN:0960-894X. (Elsevier B.V.)
      We report the use of fragment screening and fragment based drug design to develop a PI3γ kinase fragment hit into a lead. Initial fragment hits were discovered by high concn. biochem. screening, followed by a round of virtual screening to identify addnl. ligand efficient fragments. These were developed into potent and ligand efficient lead compds. using structure guided fragment growing and merging strategies. This led to a potent, selective, and cell permeable PI3γ kinase inhibitor, 12 (I), with good metabolic stability that was useful as a preclin. tool compd.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXht12jtrjO&md5=be9322ee0268040ed1da843025df4cde
    20. 20
      Brough, P. A.; Barril, X.; Borgognoni, J.; Chene, P.; Davies, N. G. M.; Davis, B.; Drysdale, M. J.; Dymock, B.; Eccles, S. A.; Garcia-Echeverria, C.; Fromont, C.; Hayes, A.; Hubbard, R. E.; Jordan, A. M.; Jensen, M. R.; Massey, A.; Merrett, A.; Padfield, A.; Parsons, R.; Radimerski, T.; Raynaud, F. I.; Robertson, A.; Roughley, S. D.; Schoepfer, J.; Simmonite, H.; Sharp, S. Y.; Surgenor, A.; Valenti, M.; Walls, S.; Webb, P.; Wood, M.; Workman, P.; Wright, L. Combining hit identification strategies: fragment-based and in silico approaches to orally active 2-aminothieno[2,3-d]pyrimidine inhibitors of the Hsp90 molecular chaperone. J. Med. Chem. 2009, 52, 4794– 4809,  DOI: 10.1021/jm900357y
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      20
      Combining Hit Identification Strategies: Fragment-Based and in Silico Approaches to Orally Active 2-Aminothieno[2,3-d]pyrimidine Inhibitors of the Hsp90 Molecular Chaperone
      Brough, Paul A.; Barril, Xavier; Borgognoni, Jenifer; Chene, Patrick; Davies, Nicholas G. M.; Davis, Ben; Drysdale, Martin J.; Dymock, Brian; Eccles, Suzanne A.; Garcia-Echeverria, Carlos; Fromont, Christophe; Hayes, Angela; Hubbard, Roderick E.; Jordan, Allan M.; Jensen, Michael Rugaard; Massey, Andrew; Merrett, Angela; Padfield, Antony; Parsons, Rachel; Radimerski, Thomas; Raynaud, Florence I.; Robertson, Alan; Roughley, Stephen D.; Schoepfer, Joseph; Simmonite, Heather; Sharp, Swee Y.; Surgenor, Allan; Valenti, Melanie; Walls, Steven; Webb, Paul; Wood, Mike; Workman, Paul; Wright, Lisa
      Journal of Medicinal Chemistry (2009), 52 (15), 4794-4809CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Inhibitors of the Hsp90 mol. chaperone are showing considerable promise as potential mol. therapeutic agents for the treatment of cancer. Here we describe novel 2-aminothieno[2,3-d]pyrimidine ATP competitive Hsp90 inhibitors, which were designed by combining structural elements of distinct low affinity hits generated from fragment-based and in silico screening exercises in concert with structural information from X-ray protein crystallog. Examples from this series have high affinity (IC50 = 50-100 nM) for Hsp90 as measured in a fluorescence polarization (FP) competitive binding assay and are active in human cancer cell lines where they inhibit cell proliferation and exhibit a characteristic profile of depletion of oncogenic proteins and concomitant elevation of Hsp72. Several compds. caused tumor growth regression at well tolerated doses when administered orally in a human BT474 human breast cancer xenograft model.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXos1ehsbs%253D&md5=4d1988060b0f87dc776e1972ea14808e
    21. 21
      Nikiforov, P. O.; Surade, S.; Blaszczyk, M.; Delorme, V.; Brodin, P.; Baulard, A. R.; Blundell, T. L.; Abell, C. A fragment merging approach towards the development of small molecule inhibitors of Mycobacterium tuberculosis EthR for use as ethionamide boosters. Org. Biomol. Chem. 2016, 14, 2318– 2326,  DOI: 10.1039/C5OB02630J
      [Crossref], [PubMed], [CAS], Google Scholar
      21
      A fragment merging approach towards the development of small molecule inhibitors of Mycobacterium tuberculosis EthR for use as ethionamide boosters
      Nikiforov, Petar O.; Surade, Sachin; Blaszczyk, Michal; Delorme, Vincent; Brodin, Priscille; Baulard, Alain R.; Blundell, Tom L.; Abell, Chris
      Organic & Biomolecular Chemistry (2016), 14 (7), 2318-2326CODEN: OBCRAK; ISSN:1477-0520. (Royal Society of Chemistry)
      With the ever-increasing instances of resistance to frontline TB drugs there is the need to develop novel strategies to fight the worldwide TB epidemic. Boosting the effect of the existing second-line antibiotic ethionamide by inhibiting the mycobacterial transcriptional repressor protein EthR is an attractive therapeutic strategy. Herein we report the use of a fragment based drug discovery approach for the structure-guided systematic merging of two fragment mols., each binding twice to the hydrophobic cavity of EthR from M. tuberculosis. These together fill the entire binding pocket of EthR. We elaborated these fragment hits and developed small mol. inhibitors which have a 100-fold improvement of potency in vitro over the initial fragments.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28Xoslajtg%253D%253D&md5=bbe5d5a96e650841b351f2fb9d221f22
    22. 22
      Schade, M.; Merla, B.; Lesch, B.; Wagener, M.; Timmermanns, S.; Pletinckx, K.; Hertrampf, T. Highly selective sub-nanomolar cathepsin S inhibitors by merging fragment binders with nitrile inhibitors. J. Med. Chem. 2020, 63, 11801– 11808,  DOI: 10.1021/acs.jmedchem.0c00949
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      22
      Highly Selective Sub-Nanomolar Cathepsin S Inhibitors by Merging Fragment Binders with Nitrile Inhibitors
      Schade, Markus; Merla, Beatrix; Lesch, Bernhard; Wagener, Markus; Timmermanns, Simone; Pletinckx, Katrien; Hertrampf, Torsten
      Journal of Medicinal Chemistry (2020), 63 (20), 11801-11808CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Pharmacol. inhibition of cathepsin S (CatS) allows for a specific modulation of the adaptive immune system and many major diseases. Here, we used NMR fragment screening and crystal structure-aided merging to synthesize novel, highly selective CatS inhibitors with picomolar enzymic Ki values and nanomolar functional activity in human Raji cells. Noncovalent fragment hits revealed binding hotspots, while the covalent inhibitor structure-activity relationship enabled efficient potency optimization.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhslGqsrnI&md5=f7698389131c792e3bebb40123f6ec66
    23. 23
      de Souza Neto, L. R.; Moreira-Filho, J. T.; Neves, B. J.; Maidana, R. L. B. R.; Guimarães, A. C. R.; Furnham, N.; Andrade, C. H.; Silva, F. P. In silico strategies to support fragment-to-lead optimization in drug discovery. Front. Chem. 2020, 8, 93,  DOI: 10.3389/fchem.2020.00093
      [Crossref], [PubMed], [CAS], Google Scholar
      23
      In silico Strategies to Support Fragment-to-Lead Optimization in Drug Discovery
      de Souza Neto Lauro Ribeiro; Maidana Rocio Lucia Beatriz Riveros; Silva Floriano Paes Jr; Moreira-Filho Jose Teofilo; Neves Bruno Junior; Andrade Carolina Horta; Neves Bruno Junior; Maidana Rocio Lucia Beatriz Riveros; Guimaraes Ana Carolina Ramos; Furnham Nicholas
      Frontiers in chemistry (2020), 8 (), 93 ISSN:2296-2646.
      Fragment-based drug (or lead) discovery (FBDD or FBLD) has developed in the last two decades to become a successful key technology in the pharmaceutical industry for early stage drug discovery and development. The FBDD strategy consists of screening low molecular weight compounds against macromolecular targets (usually proteins) of clinical relevance. These small molecular fragments can bind at one or more sites on the target and act as starting points for the development of lead compounds. In developing the fragments attractive features that can translate into compounds with favorable physical, pharmacokinetics and toxicity (ADMET-absorption, distribution, metabolism, excretion, and toxicity) properties can be integrated. Structure-enabled fragment screening campaigns use a combination of screening by a range of biophysical techniques, such as differential scanning fluorimetry, surface plasmon resonance, and thermophoresis, followed by structural characterization of fragment binding using NMR or X-ray crystallography. Structural characterization is also used in subsequent analysis for growing fragments of selected screening hits. The latest iteration of the FBDD workflow employs a high-throughput methodology of massively parallel screening by X-ray crystallography of individually soaked fragments. In this review we will outline the FBDD strategies and explore a variety of in silico approaches to support the follow-up fragment-to-lead optimization of either: growing, linking, and merging. These fragment expansion strategies include hot spot analysis, druggability prediction, SAR (structure-activity relationships) by catalog methods, application of machine learning/deep learning models for virtual screening and several de novo design methods for proposing synthesizable new compounds. Finally, we will highlight recent case studies in fragment-based drug discovery where in silico methods have successfully contributed to the development of lead compounds.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB387psFOgtg%253D%253D&md5=3fac2ef52a5f51e4e34826f08d5e28d5
    24. 24
      Pierce, A. C.; Rao, G.; Bemis, G. W. BREED:generating novel inhibitors through hybridization of known ligands. Application to CDK2, P38, and HIV protease. J. Med. Chem. 2004, 47, 2768– 2775,  DOI: 10.1021/jm030543u
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      24
      BREED: Generating Novel Inhibitors through Hybridization of Known Ligands. Application to CDK2, P38, and HIV Protease
      Pierce, Albert C.; Rao, Govinda; Bemis, Guy W.
      Journal of Medicinal Chemistry (2004), 47 (11), 2768-2775CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      In this work we describe BREED, a method for the generation of novel inhibitors from structures of known ligands bound to a common target. The method is essentially an automation of the common medicinal chem. practice of joining fragments of two known ligands to generate a new inhibitor. The ligand-bound target structures are overlaid, all overlapping bonds in all pairs of ligands are found, and the fragments on each side of each matching bond are swapped to generate the new mols. Since the method is automated, it can be applied recursively to generate all possible combinations of known ligands. In an application of this method to HIV protease inhibitors and protein kinase inhibitors, hundreds of new mol. structures were generated. These included known inhibitor scaffolds not included in the initial set, entirely novel scaffolds, and novel substituents on known scaffolds. The method is fast, and since all of the ligand functional groups are known to bind the target in the precise position and orientation present in the novel ligand, the success rate of this method should be superior to more traditional de novo design techniques. In an era of increasingly high-throughput structural biol., such methods for high-throughput utilization of structural information will become increasingly valuable.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXjt1ajurk%253D&md5=e0bec9082fab05b694742498b3f41549
    25. 25
      Lindert, S.; Durrant, J. D.; McCammon, J. A. LigMerge: a fast algorithm to generate models of novel potential ligands from sets of known binders. Chem. Biol. Drug Des. 2012, 80, 358– 365,  DOI: 10.1111/j.1747-0285.2012.01414.x
      [Crossref], [PubMed], [CAS], Google Scholar
      25
      LigMerge: a fast algorithm to generate models of novel potential ligands from sets of known binders
      Lindert, Steffen; Durrant, Jacob D.; McCammon, J. Andrew
      Chemical Biology & Drug Design (2012), 80 (3), 358-365CODEN: CBDDAL; ISSN:1747-0277. (Wiley-Blackwell)
      One common practice in drug discovery is to optimize known or suspected ligands to improve binding affinity. In performing these optimizations, it is useful to look at as many known inhibitors as possible for guidance. Medicinal chemists often seek to improve potency by altering certain chem. moieties of known/endogenous ligands while retaining those crit. for binding. To our knowledge, no automated, ligand-based algorithm exists for systematically "swapping" the chem. moieties of known ligands to generate novel ligands with potentially improved potency. To address this need, we have created a novel algorithm called "LigMerge". LigMerge identifies the max. (largest) common substructure of two three-dimensional ligand models, superimposes these two substructures, and then systematically mixes and matches the distinct fragments attached to the common substructure at each common atom, thereby generating multiple compd. models related to the known inhibitors that can be evaluated using computer docking prior to synthesis and exptl. testing. To demonstrate the utility of LigMerge, we identify compds. predicted to inhibit peroxisome proliferator-activated receptor gamma, HIV reverse transcriptase, and dihydrofolate reductase with affinities higher than those of known ligands. We hope that LigMerge will be a helpful tool for the drug design community.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XhtF2itbbF&md5=07cc659c48eccc967fc3b135182e9012
    26. 26
      Wang, H.; Pan, X.; Zhang, Y.; Wang, X.; Xiao, X.; Ji, C. MolHyb: a web server for structure-based drug design by molecular hybridization. J. Chem. Inf. Model. 2022, 62, 2916– 2922,  DOI: 10.1021/acs.jcim.2c00443
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      26
      MolHyb: A Web Server for Structure-Based Drug Design by Molecular Hybridization
      Wang, Hao; Pan, Xiaolin; Zhang, Yueqing; Wang, Xingyu; Xiao, Xudong; Ji, Changge
      Journal of Chemical Information and Modeling (2022), 62 (12), 2916-2922CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Mol. hybridization is a widely used ligand design method in drug discovery. In this study, we present MolHyb, a web server for structure-based ligand design by mol. hybridization. The input of MolHyb is a protein file and a seed compd. file. MolHyb tries to generate novel ligands through hybridizing the seed compd. with helper compds. that bind to the same protein target or similar proteins. To facilitate the job of getting helper compds., we compiled a modeled protein-ligand structure database as an extension to crystal structures in the PDB database by placing the bioactive compds. in ChEMBL into their corresponding 3D protein binding pocket properly. MolHyb works by searching for helper compds. from the protein-ligand structure database and migrating chem. moieties from helper compds. to the seed compd. efficiently. Hybridization is performed at both cyclic and acyclic bonds. The users can also input their own helper compds. to MolHyb. We hope that MolHyb will be a useful tool for rational drug design. MolHyb is freely available at http://molhyb.xundrug.cn/.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhsVyjsr3I&md5=f7682a620c9ca25bfbc1142b204e20fa
    27. 27
      Li, Y.; Zhao, Y.; Liu, Z.; Wang, R. Automatic Tailoring and Transplanting: a practical method that makes virtual screening more useful. J. Chem. Inf. Model. 2011, 51, 1474– 1491,  DOI: 10.1021/ci200036m
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      27
      Automatic Tailoring and Transplanting: A Practical Method that Makes Virtual Screening More Useful
      Li, Yan; Zhao, Yuan; Liu, Zhihai; Wang, Renxiao
      Journal of Chemical Information and Modeling (2011), 51 (6), 1474-1491CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Docking-based virtual screening of large compd. libraries has been widely applied to lead discovery in structure-based drug design. However, subsequent lead optimizations often rely on other types of computational methods, such as de novo design methods. We have developed an automatic method, namely automatic tailoring and transplanting (AutoT&T), which can effectively utilize the outcomes of virtual screening in lead optimization. This method detects suitable fragments on virtual screening hits and then transplants them onto a lead compd. to generate new ligand mols. Binding affinities, synthetic feasibilities, and drug-likeness properties are considered in the selection of final designs. In this study, our AutoT&T program was tested on three different target proteins, including p38 MAP kinase, PPAR-α, and Mcl-1. In the first two cases, AutoT&T was able to produce mols. identical or similar to known inhibitors with better potency than the given lead compd. In the third case, we demonstrated how to apply AutoT&T to design novel ligand mols. from scratch. Compared to the solns. generated by other two de novo design methods, i.e., LUDI and EA-Inventor, the solns. generated by AutoT&T were structurally more diverse and more promising in terms of binding scores in all three cases. AutoT&T also completed the assigned jobs more efficiently than LUDI and EA-Inventor by several folds. Our AutoT&T method has certain tech. advantages over de novo design methods. Importantly, it expands the application of virtual screening from lead discovery to lead optimization and thus may serve as a valuable tool for many researchers.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXlvFKqsb0%253D&md5=3327b42fecda9f2af07a6864d05afaa4
    28. 28
      Li, Y.; Zhao, Z.; Liu, Z.; Su, M.; Wang, R. AutoT&T v.2: an efficient and versatile tool for lead structure generation and optimization. J. Chem. Inf. Model. 2016, 56, 435– 453,  DOI: 10.1021/acs.jcim.5b00691
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      28
      AutoT&T v.2: An Efficient and Versatile Tool for Lead Structure Generation and Optimization
      Li, Yan; Zhao, Zhixiong; Liu, Zhihai; Su, Minyi; Wang, Renxiao
      Journal of Chemical Information and Modeling (2016), 56 (2), 435-453CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      In structure-based drug design, automated de novo design methods are helpful tools for lead discovery as well as lead optimization. In a previous study the authors reported a new de novo design method, namely, Automatic Tailoring and Transplanting (AutoT&T). It overcomes some intrinsic problems in conventional fragment-based buildup methods. In this study, the authors describe an upgraded version, namely, AutoT&T2. Structural operations conducted by AutoT&T2 have been largely optimized by introducing several new algorithms. As a result, its overall speed in multiround optimization jobs has been improved by a few thousand fold. With this improvement, it is now practical to conduct structural crossover among multiple lead mols. using AutoT&T2. Three different test cases are described in this study that demonstrate the new features and versatile applications of AutoT&T2. The AutoT&T2 software suite is available to the public. Besides, a Web portal for running AutoT&T2 online is provided at http://www.sioc-ccbg.ac.cn/software/att2 for testing.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28Xht1KgurY%253D&md5=765cef8f54f2367d902682c547abaf52
    29. 29
      Nisius, B.; Rester, U. Fragment shuffling: an automated workflow for three-dimensional fragment-based ligand design. J. Chem. Inf. Model. 2009, 49, 1211– 1222,  DOI: 10.1021/ci8004572
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      29
      Fragment Shuffling: An Automated Workflow for Three-Dimensional Fragment-Based Ligand Design
      Nisius, Britta; Rester, Ulrich
      Journal of Chemical Information and Modeling (2009), 49 (5), 1211-1222CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Fragment-based approaches display a promising alternative in lead discovery. Herein, we present the automated fragment shuffling workflow for the identification of novel lead compds. combining central elements from fragment-based lead identification and structure-based de novo design. Our method is based on sets of aligned 3D ligand structures binding to the same target or target family. The implementation comprises three different ligand fragmentation methods, a scoring scheme assigning individual scores to each fragment, and the incremental construction of novel ligands based on a greedy search algorithm guided by the calcd. fragment scores. The validation of our 3D ligand design workflow is presented on the basis of two pharmaceutically relevant drug targets. A retrospective study based on a selected protein kinase data set revealed that the fragment shuffling approach realizes extended results compared to the well-known BREED technique. Furthermore, we applied our approach in a prospective study for the design of novel non-peptidic thrombin inhibitors. The designed ligand structures in both studies demonstrate the potential of the fragment shuffling workflow.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXltlKls74%253D&md5=87aae00271c326a921cf38e7770405c8
    30. 30
      Maass, P.; Schulz-Gasch, T.; Stahl, M.; Rarey, M. Recore: a fast and versatile method for scaffold hopping based on small molecule crystal structure conformations. J. Chem. Inf. Model. 2007, 47, 390– 399,  DOI: 10.1021/ci060094h
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      30
      Recore: A Fast and Versatile Method for Scaffold Hopping Based on Small Molecule Crystal Structure Conformations
      Maass, Patrick; Schulz-Gasch, Tanja; Stahl, Martin; Rarey, Matthias
      Journal of Chemical Information and Modeling (2007), 47 (2), 390-399CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Replacing central elements of known active structures is a common procedure to enter new compd. classes. Different computational methods have already been developed to help with this task, varying in the description of possible replacements, the query input, and the similarity measure used. In this paper, a novel approach for scaffold replacement and a corresponding software tool, called Recore, is introduced. In contrast to prior methods, the authors main objective was to combine the following three properties in one tool: to avoid structures with strained conformations, to enable the exploration of large search spaces, and to allow interactive use through short response times. The authors introduce a new technique employing 3D fragments generated by combinatorial enumeration of cuts. It allows focusing on fragments suitable for scaffold replacement while retaining conformational information of the corresponding crystal structures. Based on this idea, the authors present an algorithm utilizing a geometric rank searching approach. Given a geometric arrangement of two or three exit vectors and addnl. pharmacophore features, the algorithm finds fragments fulfilling all these constraints ordered by increasing deviation from the query constraints. For the validation of the approach, three different drug design scenarios have been used. The results obtained show that the authors approach is able to propose new valid scaffold topologies.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2sXhvVWjtbk%253D&md5=e9d4e277da49f4ffe2c21472990c2387
    31. 31
      Polishchuk, P. CReM: chemically reasonable mutations framework for structure generation. J. Cheminf. 2020, 12, 28,  DOI: 10.1186/s13321-020-00431-w
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    32. 32
      Lim, J.; Hwang, S.-Y.; Moon, S.; Kim, S.; Kim, W. Y. Scaffold-based molecular design with a graph generative model. Chem. Sci. 2020, 11, 1153– 1164,  DOI: 10.1039/C9SC04503A
      [Crossref], [CAS], Google Scholar
      32
      Scaffold-based molecular design with a graph generative model
      Lim, Jaechang; Hwang, Sang-Yeon; Moon, Seokhyun; Kim, Seungsu; Kim, Woo Youn
      Chemical Science (2020), 11 (4), 1153-1164CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
      Searching for new mols. in areas like drug discovery often starts from the core structures of known mols. Such a method has called for a strategy of designing deriv. compds. retaining a particular scaffold as a substructure. On this account, our present work proposes a graph generative model that targets its use in scaffold-based mol. design. Our model accepts a mol. scaffold as input and extends it by sequentially adding atoms and bonds. The generated mols. are then guaranteed to contain the scaffold with certainty, and their properties can be controlled by conditioning the generation process on desired properties. The learned rule of extending mols. can well generalize to arbitrary kinds of scaffolds, including those unseen during learning. In the conditional generation of mols., our model can simultaneously control multiple chem. properties despite the search space constrained by fixing the substructure. As a demonstration, we applied our model to designing inhibitors of the epidermal growth factor receptor and show that our model can employ a simple semi-supervised extension to broaden its applicability to situations where only a small amt. of data is available.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXit1Ortr%252FO&md5=be36a65abcce15f18b4c1e529bffd905
    33. 33
      Imrie, F.; Hadfield, T. E.; Bradley, A. R.; Deane, C. M. Deep generative design with 3D pharmacophoric constraints. Chem. Sci. 2021, 12, 14577– 14589,  DOI: 10.1039/D1SC02436A
      [Crossref], [PubMed], [CAS], Google Scholar
      33
      Deep generative design with 3D pharmacophoric constraints
      Imrie, Fergus; Hadfield, Thomas E.; Bradley, Anthony R.; Deane, Charlotte M.
      Chemical Science (2021), 12 (43), 14577-14589CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
      Generative models have increasingly been proposed as a soln. to the mol. design problem. However, it has proved challenging to control the design process or incorporate prior knowledge, limiting their practical use in drug discovery. In particular, generative methods have made limited use of three-dimensional (3D) structural information even though this is crit. to binding. This work describes a method to incorporate such information and demonstrates the benefit of doing so. We combine an existing graph-based deep generative model, DeLinker, with a convolutional neural network to utilize phys.-meaningful 3D representations of mols. and target pharmacophores. We apply our model, DEVELOP, to both linker and R-group design, demonstrating its suitability for both hit-to-lead and lead optimization. The 3D pharmacophoric information results in improved generation and allows greater control of the design process. In multiple large-scale evaluations, we show that including 3D pharmacophoric constraints results in substantial improvements in the quality of generated mols. On a challenging test set derived from PDBbind, our model improves the proportion of generated mols. with high 3D similarity to the original mol. by over 300%. In addn., DEVELOP recovers 10x more of the original mols. compared to the baseline DeLinker method. Our approach is general-purpose, readily modifiable to alternate 3D representations, and can be incorporated into other generative frameworks.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXitlSrur%252FL&md5=1e51c4d1f28c6c8dff140c4c32d817f4
    34. 34
      Hadfield, T. E.; Imrie, F.; Merritt, A.; Birchall, K.; Deane, C. M. Incorporating target-specific pharmacophoric information into deep generative models for fragment elaboration. J. Chem. Inf. Model. 2022, 62, 2280– 2292,  DOI: 10.1021/acs.jcim.1c01311
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      34
      Incorporating Target-Specific Pharmacophoric Information into Deep Generative Models for Fragment Elaboration
      Hadfield, Thomas E.; Imrie, Fergus; Merritt, Andy; Birchall, Kristian; Deane, Charlotte M.
      Journal of Chemical Information and Modeling (2022), 62 (10), 2280-2292CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Despite recent interest in deep generative models for scaffold elaboration, their applicability to fragment-to-lead campaigns has so far been limited. This is primarily due to their inability to account for local protein structure or a user's design hypothesis. We propose a novel method for fragment elaboration, STRIFE, that overcomes these issues. STRIFE takes as input fragment hotspot maps (FHMs) extd. from a protein target and processes them to provide meaningful and interpretable structural information to its generative model, which in turn is able to rapidly generate elaborations with complementary pharmacophores to the protein. In a large-scale evaluation, STRIFE outperforms existing, structure-unaware, fragment elaboration methods in proposing highly ligand-efficient elaborations. In addn. to automatically extg. pharmacophoric information from a protein target's FHM, STRIFE optionally allows the user to specify their own design hypotheses.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhtFGru7rE&md5=c639b01f0c509bbd2e20308fa044bdb5
    35. 35
      Arús-Pous, J.; Patronov, A.; Bjerrum, E. J.; Tyrchan, C.; Reymond, J.-L.; Chen, H.; Engkvist, O. SMILES-based deep generative scaffold decorator for de novo drug design. J. Cheminf. 2020, 12, 38,  DOI: 10.1186/s13321-020-00441-8
      [Crossref], [CAS], Google Scholar
      35
      SMILES-based deep generative scaffold decorator for de-novo drug design
      Arus-Pous, Josep; Patronov, Atanas; Bjerrum, Esben Jannik; Tyrchan, Christian; Reymond, Jean-Louis; Chen, Hongming; Engkvist, Ola
      Journal of Cheminformatics (2020), 12 (1), 38CODEN: JCOHB3; ISSN:1758-2946. (SpringerOpen)
      Herein we report a new SMILES-based mol. generative architecture that generates mols. from scaffolds and can be trained from any arbitrary mol. set. This approach is possible thanks to a new mol. set pre-processing algorithm that exhaustively slices all possible combinations of acyclic bonds of every mol., combinatorically obtaining a large no. of scaffolds with their resp. decorations. Two examples showcasing the potential of the architecture in medicinal and synthetic chem. are described: First, models were trained with a training set obtained from a small set of Dopamine Receptor D2 active modulators and were able to meaningfully decorate a wide range of scaffolds and obtain mol. series predicted active on DRD2. Second, a larger set of drug-like mols. from ChEMBL was selectively sliced using synthetic chem. constraints. This filtering process allowed models trained with this dataset to selectively decorate diverse scaffolds with fragments that were generally predicted to be synthesizable and attachable to the scaffold using known synthetic approaches. In both cases, the models were already able to decorate mols. using specific knowledge without the need to add it with other techniques, such as reinforcement learning. We envision that this architecture will become a useful addn. to the already existent architectures for de novo mol. generation.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVyksLjK&md5=05a686a14f942874fd6f6ddbc85017a1
    36. 36
      Fialková, V.; Zhao, J.; Papadopoulos, K.; Engkvist, O.; Bjerrum, E. J.; Kogej, T.; Patronov, A. LibINVENT: reaction-based generative scaffold decoration for in silico library design. J. Chem. Inf. Model. 2022, 62, 2046– 2063,  DOI: 10.1021/acs.jcim.1c00469
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      36
      LibINVENT: Reaction-based Generative Scaffold Decoration for in Silico Library Design
      Fialkova, Vendy; Zhao, Jiaxi; Papadopoulos, Kostas; Engkvist, Ola; Bjerrum, Esben Jannik; Kogej, Thierry; Patronov, Atanas
      Journal of Chemical Information and Modeling (2022), 62 (9), 2046-2063CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Because of the strong relationship between the desired mol. activity and its structural core, the screening of focused, core-sharing chem. libraries is a key step in lead optimization. Despite the plethora of current research focused on in silico methods for mol. generation, to our knowledge, no tool capable of designing such libraries has been proposed. In this work, we present a novel tool for de novo drug design called LibINVENT. It is capable of rapidly proposing chem. libraries of compds. sharing the same core while maximizing a range of desirable properties. To further help the process of designing focused libraries, the user can list specific chem. reactions that can be used for the library creation. LibINVENT is therefore a flexible tool for generating virtual chem. libraries for lead optimization in a broad range of scenarios. Addnl., the shared core ensures that the compds. in the library are similar, possess desirable properties, and can also be synthesized under the same or similar conditions. The LibINVENT code is freely available in our public repository at https://github.com/MolecularAI/Lib-INVENT. The code necessary for data preprocessing is further available at: https://github.com/MolecularAI/Lib-INVENT-dataset.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvV2iu7vI&md5=dc7b0da044f6aa1c2554a43e0aa14672
    37. 37
      Li, Y.; Hu, J.; Wang, Y.; Zhou, J.; Zhang, L.; Liu, Z. DeepScaffold: a comprehensive tool for scaffold-based de novo drug discovery using deep learning. J. Chem. Inf. Model. 2020, 60, 77– 91,  DOI: 10.1021/acs.jcim.9b00727
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      37
      DeepScaffold: A Comprehensive Tool for Scaffold-Based De Novo Drug Discovery Using Deep Learning
      Li, Yibo; Hu, Jianxing; Wang, Yanxing; Zhou, Jielong; Zhang, Liangren; Liu, Zhenming
      Journal of Chemical Information and Modeling (2020), 60 (1), 77-91CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      The ultimate goal of drug design is to find novel compds. with desirable pharmacol. properties. Designing mols. retaining particular scaffolds as their core structures is an efficient way to obtain potential drug candidates. We propose a scaffold-based mol. generative model for drug discovery, which performs mol. generation based on a wide spectrum of scaffold definitions, including Bemis-Murcko scaffolds, cyclic skeletons, and scaffolds with specifications on side-chain properties. The model can generalize the learned chem. rules of adding atoms and bonds to a given scaffold. The generated compds. were evaluated by mol. docking in DRD2 targets, and the results demonstrated that this approach can be effectively applied to solve several drug design problems, including the generation of compds. contg. a given scaffold and de novo drug design of potential drug candidates with specific docking scores.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXitlerurnN&md5=44ba7010e4824a37d97c66eed5bd1d6d
    38. 38
      Yang, Y.; Zheng, S.; Su, S.; Zhao, C.; Xu, J.; Chen, H. SyntaLinker: automatic fragment linking with deep conditional transformer neural networks. Chem. Sci. 2020, 11, 8312– 8322,  DOI: 10.1039/D0SC03126G
      [Crossref], [PubMed], [CAS], Google Scholar
      38
      SyntaLinker: automatic fragment linking with deep conditional transformer neural networks
      Yang, Yuyao; Zheng, Shuangjia; Su, Shimin; Zhao, Chao; Xu, Jun; Chen, Hongming
      Chemical Science (2020), 11 (31), 8312-8322CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
      Linking fragments to generate a focused compd. library for a specific drug target is one of the challenges in fragment-based drug design (FBDD). Hereby, we propose a new program named SyntaLinker, which is based on a syntactic pattern recognition approach using deep conditional transformer neural networks. This state-of-the-art transformer can link mol. fragments automatically by learning from the knowledge of structures in medicinal chem. databases (e.g.ChEMBL database). Conventionally, linking mol. fragments was viewed as connecting substructures that were predefined by empirical rules. In SyntaLinker, however, the rules of linking fragments can be learned implicitly from known chem. structures by recognizing syntactic patterns embedded in SMILES notations. With deep conditional transformer neural networks, SyntaLinker can generate mol. structures based on a given pair of fragments and addnl. restrictions. Case studies have demonstrated the advantages and usefulness of SyntaLinker in FBDD.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhsVagsrvK&md5=403cc7a09f96ecbe27afa3ce6bee51a2
    39. 39
      Feng, Y.; Yang, Y.; Deng, W.; Chen, H.; Ran, T. SyntaLinker-Hybrid: a deep learning approach for target specific drug design. Artif. Intell. Life Sci. 2022, 2, 100035,  DOI: 10.1016/j.ailsci.2022.100035
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    40. 40
      Huang, Y.; Peng, X.; Ma, J.; Zhang, M. 3DLinker: an E(3) equivariant variational autoencoder for molecular linker design. arXiv Preprint , arXiv:2205.07309, 2022.
      Google Scholar
      There is no corresponding record for this reference.
    41. 41
      Imrie, F.; Bradley, A. R.; van der Schaar, M.; Deane, C. M. Deep generative models for 3D linker design. J. Chem. Inf. Model. 2020, 60, 1983– 1995,  DOI: 10.1021/acs.jcim.9b01120
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      41
      Deep Generative Models for 3D Linker Design
      Imrie, Fergus; Bradley, Anthony R.; van der Schaar, Mihaela; Deane, Charlotte M.
      Journal of Chemical Information and Modeling (2020), 60 (4), 1983-1995CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      Rational compd. design remains a challenging problem for both computational methods and medicinal chemists. Computational generative methods have begun to show promising results for the design problem. However, they have not yet used the power of three-dimensional (3D) structural information. We have developed a novel graph-based deep generative model that combines state-of-the-art machine learning techniques with structural knowledge. Our method ("DeLinker") takes two fragments or partial structures and designs a mol. incorporating both. The generation process is protein-context-dependent, utilizing the relative distance and orientation between the partial structures. This 3D information is vital to successful compd. design, and we demonstrate its impact on the generation process and the limitations of omitting such information. In a large-scale evaluation, DeLinker designed 60% more mols. with high 3D similarity to the original mol. than a database baseline. When considering the more relevant problem of longer linkers with at least five atoms, the outperformance increased to 200%. We demonstrate the effectiveness and applicability of this approach on a diverse range of design problems: fragment linking, scaffold hopping, and proteolysis targeting chimera (PROTAC) design. As far as we are aware, this is the first mol. generative model to incorporate 3D structural information directly in the design process. The code is available at https://github.com/oxpig/DeLinker.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXltlenurs%253D&md5=430c929761c5020d1013538e028ba1bc
    42. 42
      Andrianov, G. V.; Gabriel Ong, W. J.; Serebriiskii, I.; Karanicolas, J. Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging. J. Chem. Inf. Model. 2021, 61, 5967– 5987,  DOI: 10.1021/acs.jcim.1c00630
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      42
      Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging
      Andrianov, Grigorii V.; Gabriel Ong, Wern Juin; Serebriiskii, Ilya; Karanicolas, John
      Journal of Chemical Information and Modeling (2021), 61 (12), 5967-5987CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      In early-stage drug discovery, the hit-to-lead optimization (or "hit expansion") stage entails starting from a newly identified active compd. and improving its potency or other properties. Traditionally, this process relies on synthesizing and evaluating a series of analogs to build up structure-activity relationships. Here, we describe a computational strategy focused on kinase inhibitors, intended to expedite the process of identifying analogs with improved potency. Our protocol begins from an inhibitor of the target kinase and generalizes the synthetic route used to access it. By searching for com. available replacements for the individual building blocks used to make the parent inhibitor, we compile an enumerated library of compds. that can be accessed using the same chem. transformations; these huge libraries can exceed many millions-or billions-of compds. Because the resulting libraries are much too large for explicit virtual screening, we instead consider alternate approaches to identify the top-scoring compds. We find that contributions from individual substituents are well described by a pairwise additivity approxn., provided that the corresponding fragments position their shared core in precisely the same way relative to the binding site. This key insight allows us to det. which fragments are suitable for merging into single new compds. and which are not. Further, the use of pairwise approxn. allows interaction energies to be assigned to each compd. in the library without the need for any further structure-based modeling: interaction energies instead can be reliably estd. from the energies of the component fragments, and the reduced computational requirements allow for flexible energy minimizations that allow the kinase to respond to each substitution. We demonstrate this protocol using libraries built from six representative kinase inhibitors drawn from the literature, which target five different kinases: CDK9, CHK1, CDK2, EGFRT790M, and ACK1. In each example, the enumerated library includes addnl. analogs reported by the original study to have activity, and these analogs are successfully prioritized within the library. We envision that the insights from this work can facilitate the rapid assembly and screening of increasingly large libraries for focused hit-to-lead optimization. To enable adoption of these methods and to encourage further analyses, we disseminate the computational tools needed to deploy this protocol.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisVGrs7zE&md5=d00fcbed94deb5955f19c3ee8d26af39
    43. 43
      Hall, R. J.; Murray, C. W.; Verdonk, M. L. The Fragment Network: a chemistry recommendation engine built using a graph database. J. Med. Chem. 2017, 60, 6440– 6450,  DOI: 10.1021/acs.jmedchem.7b00809
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      43
      The Fragment Network: A Chemistry Recommendation Engine Built Using a Graph Database
      Hall, Richard J.; Murray, Christopher W.; Verdonk, Marcel L.
      Journal of Medicinal Chemistry (2017), 60 (14), 6440-6450CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      The hit validation stage of a fragment-based drug discovery campaign involves probing the SAR around one or more fragment hits. This often requires a search for similar compds. in a corporate collection or from com. suppliers. The Fragment Network is a graph database that allows a user to efficiently search chem. space around a compd. of interest. The result set is chem. intuitive, naturally grouped by substitution pattern and meaningfully sorted according to the no. of observations of each transformation in medicinal chem. databases. This paper describes the algorithms used to construct and search the Fragment Network and provides examples of how it may be used in a drug discovery context.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXhtFKlsr3J&md5=79842771a068cdaabc2f9a841d791c1c
    44. 44
      Landrum, G. RDKit: A software suite for cheminformatics, computational chemistry, and predictive modeling. https://www.rdkit.org/ (accessed November 2022).
      Google Scholar
      There is no corresponding record for this reference.
    45. 45
      Neo4j, Graph Modeling Guidelines. https://neo4j.com/developer/guide-data-modeling/ (accessed November 2022).
      Google Scholar
      There is no corresponding record for this reference.
    46. 46
      Newman, J. A.; Douangamath, A.; Yadzani, S.; Yosaatmadja, Y.; Aimon, A.; Brandão-Neto, J.; Dunnett, L.; Gorrie-stone, T.; Skyner, R.; Fearon, D.; Schapira, M.; von Delft, F.; Gileadi, O. Structure, mechanism and crystallographic fragment screening of the SARS-CoV-2 NSP13 helicase. Nat. Commun. 2021, 12, 4848,  DOI: 10.1038/s41467-021-25166-6
      [Crossref], [PubMed], [CAS], Google Scholar
      46
      Structure, mechanism and crystallographic fragment screening of the SARS-CoV-2 NSP13 helicase
      Newman, Joseph A.; Douangamath, Alice; Yadzani, Setayesh; Yosaatmadja, Yuliana; Aimon, Antony; Brandao-Neto, Jose; Dunnett, Louise; Gorrie-stone, Tyler; Skyner, Rachael; Fearon, Daren; Schapira, Matthieu; von Delft, Frank; Gileadi, Opher
      Nature Communications (2021), 12 (1), 4848CODEN: NCAOBW; ISSN:2041-1723. (Nature Research)
      Abstr.: There is currently a lack of effective drugs to treat people infected with SARS-CoV-2, the cause of the global COVID-19 pandemic. The SARS-CoV-2 Non-structural protein 13 (NSP13) has been identified as a target for anti-virals due to its high sequence conservation and essential role in viral replication. Structural anal. reveals two "druggable" pockets on NSP13 that are among the most conserved sites in the entire SARS-CoV-2 proteome. Here we present crystal structures of SARS-CoV-2 NSP13 solved in the APO form and in the presence of both phosphate and a non-hydrolysable ATP analog. Comparisons of these structures reveal details of conformational changes that provide insights into the helicase mechanism and possible modes of inhibition. To identify starting points for drug development we have performed a crystallog. fragment screen against NSP13. The screen reveals 65 fragment hits across 52 datasets opening the way to structure guided development of novel antiviral agents.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhslOns7%252FI&md5=3e643800631b8bb56328c4ddd97fb7a7
    47. 47
      Jin, Z.; Du, X.; Xu, Y.; Deng, Y.; Liu, M.; Zhao, Y.; Zhang, B.; Li, X.; Zhang, L.; Peng, C.; Duan, Y.; Yu, J.; Wang, L.; Yang, K.; Liu, F.; Jiang, R.; Yang, X.; You, T.; Liu, X.; Yang, X.; Bai, F.; Liu, H.; Liu, X.; Guddat, L. W.; Xu, W.; Xiao, G.; Qin, C.; Shi, Z.; Jiang, H.; Rao, Z.; Yang, H. Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors. Nature 2020, 582, 289– 293,  DOI: 10.1038/s41586-020-2223-y
      [Crossref], [PubMed], [CAS], Google Scholar
      47
      Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors
      Jin, Zhenming; Du, Xiaoyu; Xu, Yechun; Deng, Yongqiang; Liu, Meiqin; Zhao, Yao; Zhang, Bing; Li, Xiaofeng; Zhang, Leike; Peng, Chao; Duan, Yinkai; Yu, Jing; Wang, Lin; Yang, Kailin; Liu, Fengjiang; Jiang, Rendi; Yang, Xinglou; You, Tian; Liu, Xiaoce; Yang, Xiuna; Bai, Fang; Liu, Hong; Liu, Xiang; Guddat, Luke W.; Xu, Wenqing; Xiao, Gengfu; Qin, Chengfeng; Shi, Zhengli; Jiang, Hualiang; Rao, Zihe; Yang, Haitao
      Nature (London, United Kingdom) (2020), 582 (7811), 289-293CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
      Abstr.: A new coronavirus, known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is the etiol. agent responsible for the 2019-2020 viral pneumonia outbreak of coronavirus disease 2019 (COVID-19). Currently, there are no targeted therapeutic agents for the treatment of this disease, and effective treatment options remain very limited. Here, we describe the results of a program that aimed to rapidly discover lead compds. for clin. use, by combining structure-assisted drug design, virtual drug screening and high-throughput screening. This program focused on identifying drug leads that target main protease (Mpro) of SARS-CoV-2: Mpro is a key enzyme of coronaviruses and has a pivotal role in mediating viral replication and transcription, making it an attractive drug target for SARS-CoV-2. We identified a mechanism-based inhibitor (N3) by computer-aided drug design, and then detd. the crystal structure of Mpro of SARS-CoV-2 in complex with this compd. Through a combination of structure-based virtual and high-throughput screening, we assayed more than 10,000 compds.-including approved drugs, drug candidates in clin. trials and other pharmacol. active compds.-as inhibitors of Mpro. Six of these compds. inhibited Mpro, showing half-maximal inhibitory concn. values that ranged from 0.67 to 21.4μM. One of these compds. (ebselen) also exhibited promising antiviral activity in cell-based assays. Our results demonstrate the efficacy of our screening strategy, which can lead to the rapid discovery of drug leads with clin. potential in response to new infectious diseases for which no specific drugs or vaccines are available.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVyhsrrO&md5=b84f350fe9ce1109485df6caf814ba82
    48. 48
      The COVID Moonshot Consortium. Achdout, H.; Aimon, A.; Bar-David, E.; Barr, H.; Ben-Shmuel, A.; Bennett, J.; Bilenko, V. A.; Bilenko, V. A.; Boby, M. L.; Borden, B.; Bowman, G. R.; Brun, J.; BVNBS, S.; Calmiano, M.; Carbery, A.; Carney, D.; Cattermole, E.; Chang, E.; Chernyshenko, E.; Chodera, J. D.; Clyde, A.; Coffland, J. E.; Cohen, G.; Cole, J.; Contini, A.; Cox, L.; Cvitkovic, M.; Dias, A.; Donckers, K.; Dotson, D. L.; Douangamath, A.; Duberstein, S.; Dudgeon, T.; Dunnett, L.; Eastman, P. K.; Erez, N.; Eyermann, C. J.; Fairhead, M.; Fate, G.; Fearon, D.; Fedorov, O.; Ferla, M.; Fernandes, R. S.; Ferrins, L.; Foster, R.; Foster, H.; Gabizon, R.; Garcia-Sastre, A.; Gawriljuk, V. O.; Gehrtz, P.; Gileadi, C.; Giroud, C.; Glass, W. G.; Glen, R.; Glinert, I.; Godoy, A. S.; Gorichko, M.; Gorrie-Stone, T.; Griffen, E. J.; Hart, S. H.; Heer, J.; Henry, M.; Hill, M.; Horrell, S.; Huliak, V. D.; Hurley, M. F.; Israely, T.; Jajack, A.; Jansen, J.; Jnoff, E.; Jochmans, D.; John, T.; Jonghe, S. D.; Kantsadi, A. L.; Kenny, P. W.; Kiappes, J. L.; Kinakh, S. O.; Koekemoer, L.; Kovar, B.; Krojer, T.; Lee, A.; Lefker, B. A.; Levy, H.; Logvinenko, I. G.; London, N.; Lukacik, P.; Macdonald, H. B.; MacLean, B.; Malla, T. R.; Matviiuk, T.; McCorkindale, W.; McGovern, B. L.; Melamed, S.; Melnykov, K. P.; Michurin, O.; Mikolajek, H.; Milne, B. F.; Morris, A.; Morris, G. M.; Morwitzer, M. J.; Moustakas, D.; Nakamura, A. M.; Neto, J. B.; Neyts, J.; Nguyen, L.; Noske, G. D.; Oleinikovas, V.; Oliva, G.; Overheul, G. J.; Owen, D.; Pai, R.; Pan, J.; Paran, N.; Perry, B.; Pingle, M.; Pinjari, J.; Politi, B.; Powell, A.; Psenak, V.; Puni, R.; Rangel, V. L.; Reddi, R. N.; Reid, S. P.; Resnick, E.; Ripka, E. G.; Robinson, M. C.; Robinson, R. P.; Rodriguez-Guerra, J.; Rosales, R.; Rufa, D.; Saar, K.; Saikatendu, K. S.; Schofield, C.; Shafeev, M.; Shaikh, A.; Shi, J.; Shurrush, K.; Singh, S.; Sittner, A.; Skyner, R.; Smalley, A.; Smeets, B.; Smilova, M. D.; Solmesky, L. J.; Spencer, J.; Strain-Damerell, C.; Swamy, V.; Tamir, H.; Tennant, R.; Thompson, W.; Thompson, A.; Tomasio, S.; Tsurupa, I. S.; Tumber, A.; Vakonakis, I.; van Rij, R. P.; Vangeel, L.; Varghese, F. S.; Vaschetto, M.; Vitner, E. B.; Voelz, V.; Volkamer, A.; von Delft, F.; von Delft, A.; Walsh, M.; Ward, W.; Weatherall, C.; Weiss, S.; White, K. M.; Wild, C. F.; Wittmann, M.; Wright, N.; Yahalom-Ronen, Y.; Zaidmann, D.; Zidane, H.; Zitzmann, N. Open science discovery of oral non-covalent SARS-CoV-2 main protease inhibitor therapeutics. bioRxiv Preprint , updated version, 2022.  DOI: 10.1101/2020.10.29.339317
      [Crossref], Google Scholar
      There is no corresponding record for this reference.
    49. 49
      Wahlberg, E.; Karlberg, T.; Kouznetsova, E.; Markova, N.; Macchiarulo, A.; Thorsell, A.-G.; Pol, E.; Frostell, s.; Ekblad, T.; Öncü, D.; Kull, B.; Robertson, G. M.; Pellicciari, R.; Schüler, H.; Weigelt, J. Family-wide chemical profiling and structural analysis of PARP and tankyrase inhibitors. Nat. Biotechnol. 2012, 30, 283– 288,  DOI: 10.1038/nbt.2121
      [Crossref], [PubMed], [CAS], Google Scholar
      49
      Family-wide chemical profiling and structural analysis of PARP and tankyrase inhibitors
      Wahlberg, Elisabet; Karlberg, Tobias; Kouznetsova, Ekaterina; Markova, Natalia; Macchiarulo, Antonio; Thorsell, Ann-Gerd; Pol, Ewa; Frostell, Aasa; Ekblad, Torun; Oencue, Delal; Kull, Bjoern; Robertson, Graeme Michael; Pellicciari, Roberto; Schueler, Herwig; Weigelt, Johan
      Nature Biotechnology (2012), 30 (3), 283-288CODEN: NABIF9; ISSN:1087-0156. (Nature Publishing Group)
      Inhibitors of poly-ADP-ribose polymerase (PARP) family proteins are currently in clin. trials as cancer therapeutics, yet the specificity of many of these compds. is unknown. Here we evaluated a series of 185 small-mol. inhibitors, including research reagents and compds. being tested clin., for the ability to bind to the catalytic domains of 13 of the 17 human PARP family members including the tankyrases, TNKS1 and TNKS2. Many of the best-known inhibitors, including TIQ-A, 6(5H)-phenanthridinone, olaparib, ABT-888 and rucaparib, bound to several PARP family members, suggesting that these mols. lack specificity and have promiscuous inhibitory activity. We also detd. X-ray crystal structures for five TNKS2 ligand complexes and four PARP14 ligand complexes. In addn. to showing that the majority of PARP inhibitors bind multiple targets, these results provide insight into the design of new inhibitors.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XisVKrsb4%253D&md5=71f17fd5b6bb602dd70c8fcc17df95f9
    50. 50
      Ohara-Nemoto, Y.; Shimoyama, Y.; Kimura, S.; Kon, A.; Haraga, H.; Ono, T.; Nemoto, T. K. Asp- and Glu-specific novel dipeptidyl peptidase 11 of Porphyromonas gingivalis ensures utilization of proteinaceous energy sources. J. Biol. Chem. 2011, 286, 38115– 38127,  DOI: 10.1074/jbc.M111.278572
      [Crossref], [PubMed], [CAS], Google Scholar
      50
      Asp- and Glu-specific Novel Dipeptidyl Peptidase 11 of Porphyromonas gingivalis Ensures Utilization of Proteinaceous Energy Sources
      Ohara-Nemoto, Yuko; Shimoyama, Yu; Kimura, Shigenobu; Kon, Asako; Haraga, Hiroshi; Ono, Toshio; Nemoto, Takayuki K.
      Journal of Biological Chemistry (2011), 286 (44), 38115-38127, S38115/1-S38115/9CODEN: JBCHA3; ISSN:0021-9258. (American Society for Biochemistry and Molecular Biology)
      Porphyromonas gingivalis and Porphyromonas endodontalis, asaccharolytic black-pigmented anaerobes, are predominant pathogens of human chronic and periapical periodontitis, resp. They incorporate di- and tripeptides from the environment as carbon and energy sources. In the present study we cloned a novel dipeptidyl peptidase (DPP) gene of P. endodontalis ATCC 35406, designated as DPP11. The DPP11 gene encoded 717 amino acids with a mol. mass of 81,090 Da and was present as a 75-kDa form with an N terminus of Asp22. A homol. search revealed the presence of a P. gingivalis ortholog, PGN0607, that has been categorized as an isoform of authentic DPP7. P. gingivalis DPP11 was exclusively cell-assocd. as a truncated 60-kDa form, and the gene ablation retarded cell growth. DPP11 specifically removed dipeptides from oligopeptides with the penultimate N-terminal Asp and Glu and has a P2-position preference to hydrophobic residues. Optimum pH was 7.0, and the kcat/Km value was higher for Asp than Glu. Those activities were lost by substitution of Ser652 in P. endodontalis and Ser655 in P. gingivalis DPP11 to Ala, and they were consistently decreased with increasing NaCl concn. Arg670 is a unique amino acid completely conserved in all DPP11 members distributed in the genera Porphyromonas, Bacteroides, and Parabacteroides, whereas this residue is converted to Gly in all authentic DPP7 members. Substitution anal. suggested that Arg670 interacts with an acidic residue of the substrate. Considered to preferentially utilize acidic amino acids, DPP11 ensures efficient degrdn. of oligopeptide substrates in these Gram-neg. anaerobic rods.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtlyktb%252FK&md5=419046327fe3bfdfb5619d722d9f9755
    51. 51
      Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Delivery Rev. 1997, 23, 3– 25,  DOI: 10.1016/S0169-409X(96)00423-1
      [Crossref], [CAS], Google Scholar
      51
      Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings
      Lipinski, Christopher A.; Lombardo, Franco; Dominy, Beryl W.; Feeney, Paul J.
      Advanced Drug Delivery Reviews (1997), 23 (1-3), 3-25CODEN: ADDREP; ISSN:0169-409X. (Elsevier)
      A review with 50 refs. Exptl. and computational approaches to est. soly. and permeability in discovery and development settings are described. In the discovery setting 'the rule of 5' predicts that poor absorption or permeation is more likely when there are more than 5 H-bond donors, 10 H-bond acceptors, the mol. wt. (MWT) is >500 and the calcd. Log P (CLogP) is >5. Computational methodol. for the rule-based Moriguchi Log P (MLogP) calcn. is described. Turbidimetric soly. measurement is described and applied to known drugs. High throughput screening (HTS) leads tend to have higher MWT and Log P and lower turbidimetric soly. than leads in the pre-HTS era. In the development setting, soly. calcns. focus on exact value prediction and are difficult because of polymorphism. Recent work on linear free energy relationships and Log P approaches are critically reviewed. Useful predictions are possible in closely related analog series when coupled with exptl. thermodn. soly. measurements.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2sXktlKlsQ%253D%253D&md5=405f70b0594d428f1275e1d56642cd3a
    52. 52
      Veber, D. F.; Johnson, S. R.; Cheng, H.-Y.; Smith, B. R.; Ward, K. W.; Kopple, K. D. Molecular properties that influence the oral bioavailability of drug candidates. J. Med. Chem. 2002, 45, 2615– 2623,  DOI: 10.1021/jm020017n
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      52
      Molecular Properties That Influence the Oral Bioavailability of Drug Candidates
      Veber, Daniel F.; Johnson, Stephen R.; Cheng, Hung-Yuan; Smith, Brian R.; Ward, Keith W.; Kopple, Kenneth D.
      Journal of Medicinal Chemistry (2002), 45 (12), 2615-2623CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Oral bioavailability measurements in rats for over 1100 drug candidates studied at Smith-Kline Beecham Pharmaceuticals (now Glaxo Smith-Kline) have allowed us to analyze the relative importance of mol. properties considered to influence that drug property. Reduced mol. flexibility, as measured by the no. of rotatable bonds, and low polar surface area or total hydrogen bond count (sum of donors and acceptors) are found to be important predictors of good oral bioavailability, independent of mol. wt. That on av. both the no. of rotatable bonds and polar surface area or hydrogen bond count tend to increase with mol. wt. may in part explain the success of the mol. wt. parameter in predicting oral bioavailability. The commonly applied mol. wt. cutoff at 500 does not itself significantly sep. compds. with poor oral bioavailability from those with acceptable values in this extensive data set. Our observations suggest that compds. which meet only the 2 criteria of (1) 10 or fewer rotatable bonds and (2) polar surface area ≤140 Å2 (or 12 or fewer H-bond donors and acceptors) will have a high probability of good oral bioavailability in the rat. Data sets for the artificial membrane permeation rate and for clearance in the rat were also examd. Reduced polar surface area correlates better with increased permeation rate than does lipophilicity (C log P), and increased rotatable bond count has a neg. effect on the permeation rate. A threshold permeation rate is a prerequisite of oral bioavailability. The rotatable bond count does not correlate with the data examd. here for the in vivo clearance rate in the rat.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38XjsFCmt7g%253D&md5=eaad26ed6a259de82ad65a8834fc397d
    53. 53
      Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 2012, 40, D1100– D1107,  DOI: 10.1093/nar/gkr777
      [Crossref], [PubMed], [CAS], Google Scholar
      53
      ChEMBL: a large-scale bioactivity database for drug discovery
      Gaulton, Anna; Bellis, Louisa J.; Bento, A. Patricia; Chambers, Jon; Davies, Mark; Hersey, Anne; Light, Yvonne; McGlinchey, Shaun; Michalovich, David; Al-Lazikani, Bissan; Overington, John P.
      Nucleic Acids Research (2012), 40 (D1), D1100-D1107CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
      ChEMBL is an Open Data database contg. binding, functional and ADMET information for a large no. of drug-like bioactive compds. These data are manually abstracted from the primary published literature on a regular basis, then further curated and standardized to maximize their quality and utility across a wide range of chem. biol. and drug-discovery research problems. Currently, the database contains 5.4 million bioactivity measurements for more than 1 million compds. and 5200 protein targets. Access is available through a web-based interface, data downloads and web services at: https://www.ebi.ac.uk/chembldb.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs12htbjN&md5=aedf7793e1ca54b6a4fa272ea3ef7d0e
    54. 54
      Ebejer, J.-P.; Morris, G. M.; Deane, C. M. Freely available conformer generation methods: how good are they?. J. Chem. Inf. Model. 2012, 52, 1146– 1158,  DOI: 10.1021/ci2004658
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      54
      Freely Available Conformer Generation Methods: How Good Are They?
      Ebejer, Jean-Paul; Morris, Garrett M.; Deane, Charlotte M.
      Journal of Chemical Information and Modeling (2012), 52 (5), 1146-1158CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      A review. Conformer generation has important implications in cheminformatics, particularly in computational drug discovery where the quality of conformer generation software may affect the outcome of a virtual screening exercise. We examine the performance of four freely available small mol. conformer generation tools (Balloon, Confab, Frog2, and RDKit) alongside a com. tool (MOE). The aim of this study is 3-fold: (i) to identify which tools most accurately reproduce exptl. detd. structures; (ii) to examine the diversity of the generated conformational set; and (iii) to benchmark the computational time expended. These aspects were tested using a set of 708 drug-like mols. assembled from the OMEGA validation set and the Astex Diverse Set. These mols. have varying physicochem. properties and at least one known X-ray crystal structure. We found that RDKit and Confab are statistically better than other methods at generating low rmsd conformers to the known structure. RDKit is particularly suited for less flexible mols. while Confab, with its systematic approach, is able to generate conformers which are geometrically closer to the exptl. detd. structure for mols. with a large no. of rotatable bonds (≥10). In our tests RDKit also resulted as the second fastest method after Frog2. In order to enhance the performance of RDKit, we developed a postprocessing algorithm to build a diverse and representative set of conformers which also contains a close conformer to the known structure. Our anal. indicates that, with postprocessing, RDKit is a valid free alternative to com., proprietary software.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38Xlt1Snu7s%253D&md5=2990e7dcfc83d8c0f7671250c51aeb97
    55. 55
      Su, M.; Yang, Q.; Du, Y.; Feng, G.; Liu, Z.; Li, Y.; Wang, R. Comparative assessment of scoring functions: the CASF-2016 update. J. Chem. Inf. Model. 2019, 59, 895– 913,  DOI: 10.1021/acs.jcim.8b00545
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      55
      Comparative Assessment of Scoring Functions: The CASF-2016 Update
      Su, Minyi; Yang, Qifan; Du, Yu; Feng, Guoqin; Liu, Zhihai; Li, Yan; Wang, Renxiao
      Journal of Chemical Information and Modeling (2019), 59 (2), 895-913CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
      In structure-based drug design, scoring functions are often employed to evaluate protein-ligand interactions. A variety of scoring functions have been developed so far, and thus, some objective benchmarks are desired for assessing their strength and weakness. The comparative assessment of scoring functions (CASF) benchmark developed by us provides an answer to this demand. CASF is designed as a "scoring benchmark", where the scoring process is decoupled from the docking process to depict the performance of scoring function more precisely. Here, we describe the latest update of this benchmark, i.e., CASF-2016. Each scoring function is still evaluated by four metrics, including "scoring power", "ranking power", "docking power", and "screening power". Nevertheless, the evaluation methods have been improved considerably in several aspects. A new test set is compiled, which consists of 285 protein-ligand complexes with high-quality crystal structures and reliable binding consts. A panel of 25 scoring functions are tested on CASF-2016 as a demonstration. Our results reveal that the performance of current scoring functions is more promising in terms of docking power than scoring, ranking, and screening power. Scoring power is somewhat correlated with ranking power, so are docking power and screening power. The results obtained on CASF-2016 may provide valuable guidance for the end users to make smart choices among available scoring functions. Moreover, CASF is created as an open-access benchmark so that other researchers can utilize it to test a wider range of scoring functions. The complete CASF-2016 benchmark will be released on the PDBbind-CN web server (http://www.pdbbind-cn.org/casf.asp/) once this article is published.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXitlWhtLjM&md5=573295cfd2298a2dee933027041c3d9c
    56. 56
      Ferla, M. Fragmenstein. https://github.com/matteoferla/Fragmenstein (accessed November 2022).
      Google Scholar
      There is no corresponding record for this reference.
    57. 57
      Chaudhury, S.; Lyskov, S.; Gray, J. J. PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta. Bioinformatics 2010, 26, 689– 691,  DOI: 10.1093/bioinformatics/btq007
      [Crossref], [PubMed], [CAS], Google Scholar
      57
      PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta
      Chaudhury, Sidhartha; Lyskov, Sergey; Gray, Jeffrey J.
      Bioinformatics (2010), 26 (5), 689-691CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)
      Summary: PyRosetta is a stand-alone Python-based implementation of the Rosetta mol. modeling package that allows users to write custom structure prediction and design algorithms using the major Rosetta sampling and scoring functions. PyRosetta contains Python bindings to libraries that define Rosetta functions including those for accessing and manipulating protein structure, calcg. energies and running Monte Carlo-based simulations. PyRosetta can be used in two ways: (i) interactively, using iPython and (ii) script-based, using Python scripting. Interactive mode contains a no. of help features and is ideal for beginners while script-mode is best suited for algorithm development. PyRosetta has similar computational performance to Rosetta, can be easily scaled up for cluster applications and has been implemented for algorithms demonstrating protein docking, protein folding, loop modeling and design. Availability: PyRosetta is a stand-alone package available at http://www.pyrosetta.org under the Rosetta license which is free for academic and non-profit users. A tutorial, user's manual and sample scripts demonstrating usage are also available on the web site.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXis1Wnt70%253D&md5=261b3b6b59e4987911556ea5314d9b19
    58. 58
      Adasme, M. F.; Linnemann, K. L.; Bolz, S. N.; Kaiser, F.; Salentin, S.; Haupt, V.; Schroeder, M. PLIP 2021: expanding the scope of the protein–ligand interaction profiler to DNA and RNA. Nucleic Acids Res. 2021, 49, W530– W534,  DOI: 10.1093/nar/gkab294
      [Crossref], [PubMed], [CAS], Google Scholar
      58
      PLIP 2021 expanding scope of protein-ligand interaction profiler to DNA and RNA
      Adasme, Melissa F.; Linnemann, Katja L.; Bolz, Sarah Naomi; Kaiser, Florian; Salentin, Sebastian; Haupt, V. Joachim; Schroeder, Michael
      Nucleic Acids Research (2021), 49 (W1), W530-W534CODEN: NARHAD; ISSN:1362-4962. (Oxford University Press)
      With the growth of protein structure data, the anal. of mol. interactions between ligands and their target mols. is gaining importance. PLIP, the protein-ligand interaction profiler, detects and visualises these interactions and provides data in formats suitable for further processing. PLIP has proven very successful in applications ranging from the characterization of docking expts. to the assessment of novel ligand-protein complexes. Besides ligand-protein interactions, interactions with DNA and RNA play a vital role in many applications, such as drugs targeting DNA or RNA-binding proteins. To date, over 7% of all 3D structures in the Protein Data Bank include DNA or RNA. Therefore, we extended PLIP to encompass these important mols. We demonstrate the power of this extension with examples of a cancer drug binding to a DNA target, and an RNA-protein complex central to a neurol. disease.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhvV2isLbL&md5=8f56ced733405aaa8f5325e3fa4b862c
    59. 59
      Carbery, A.; Skyner, R.; von Delft, F.; Deane, C. M. Fragment libraries designed to be functionally diverse recover protein binding information more efficiently than standard structurally diverse libraries. J. Med. Chem. 2022, 65, 11404– 11413,  DOI: 10.1021/acs.jmedchem.2c01004
      [ACS Full Text ACS Full Text], [CAS], Google Scholar
      59
      Fragment Libraries Designed to Be Functionally Diverse Recover Protein Binding Information More Efficiently Than Standard Structurally Diverse Libraries
      Carbery, Anna; Skyner, Rachael; von Delft, Frank; Deane, Charlotte M.
      Journal of Medicinal Chemistry (2022), 65 (16), 11404-11413CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
      Current fragment-based drug design relies on the efficient exploration of chem. space by using structurally diverse libraries of small fragments. However, structurally dissimilar compds. can exploit the same interactions and thus be functionally similar. Using three-dimensional structures of many fragments bound to multiple targets, we examd. if a better strategy for selecting fragments for screening libraries exists. We show that structurally diverse fragments can be described as functionally redundant, often making the same interactions. Ranking fragments by the no. of novel interactions they made, we show that functionally diverse selections of fragments substantially increase the amt. of information recovered for unseen targets compared to the amts. recovered by other methods of selection. Using these results, we design small functionally efficient libraries that can give significantly more information about new protein targets than similarly sized structurally diverse libraries. By covering more functional space, we can generate more diverse sets of drug leads.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XitFWktLnO&md5=20c210891befe38eb6ec658cba2a01b8
    60. 60
      Maaten, L. v. d.; Hinton, G. Visualizing data using t-SNE. J. Mach. Learn. Res. 2008, 9, 2579– 2605
      Google Scholar
      There is no corresponding record for this reference.
    61. 61
      Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V.; Vanderplas, J.; Passos, A.; Cournapeau, D.; Brucher, M.; Perrot, M.; Duchesnay, d. Scikit-learn: machine learning in Python. J. Mach. Learn. Res. 2011, 12, 2825– 2830
      Google Scholar
      There is no corresponding record for this reference.
    62. 62
      PostEra, Mpro activity data. https://covid.postera.ai/covid/activity_data (accessed November 2022).
      Google Scholar
      There is no corresponding record for this reference.
    63. 63
      Butina, D. Unsupervised data base clustering based on Daylight’s fingerprint and Tanimoto similarity: a fast and automated way to cluster small and large data sets. JCICS 1999, 39, 747– 750,  DOI: 10.1021/ci9803381
      [ACS Full Text ACS Full Text], Google Scholar
      There is no corresponding record for this reference.

  • Supporting Information

    Supporting Information

    ARTICLE SECTIONS
    Jump To

    The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.3c00276.

    • Figures and tables referenced in the text (PDF)


    Terms & Conditions

    Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.

PDF [ 8MB]

Pair your accounts.

Export articles to Mendeley

Get article recommendations from ACS based on references in your Mendeley library.

Pair your accounts.

Export articles to Mendeley

Get article recommendations from ACS based on references in your Mendeley library.

You’ve supercharged your research process with ACS and Mendeley!

STEP 1:
Click to create an ACS ID

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

MENDELEY PAIRING EXPIRED
Your Mendeley pairing has expired. Please reconnect