Download Hi-Res ImageDownload to MS-PowerPointCite This:J. Med. Chem. 2012, 55, 14, 6582-6594

Directory of Useful Decoys, Enhanced (DUD-E): Better Ligands and Decoys for Better Benchmarking

View Author Information

† Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, California 94158-2330, United States

*For J.J.I.: phone, (415) 514-4127; E-mail, [email protected]. For B.K.S.: phone, (415) 514-4126; E-mail, [email protected]. Address: John J. Irwin or Brian K. Shoichet, Department of Pharmaceutical Chemistry, University of California San Francisco, 1700 Fourth Street, Box 2550, San Francisco, CA 94158-2330.

Cite this: J. Med. Chem. 2012, 55, 14, 6582–6594

Publication Date (Web):June 20, 2012

https://doi.org/10.1021/jm300687e

Request reuse permissions

Article Views

31754

Altmetric

Citations

1324

LEARN ABOUT THESE METRICS

Article Views are the COUNTER-compliant sum of full text article downloads since November 2008 (both PDF and HTML) across all institutions and individuals. These metrics are regularly updated to reflect usage leading up to the last few days.

Citations are the number of other articles citing this article, calculated by Crossref and updated daily. Find more information about Crossref citation counts.

The Altmetric Attention Score is a quantitative measure of the attention that a research article has received online. Clicking on the donut icon will load a page at altmetric.com with additional details about the score and the social media presence for the given article. Find more information on the Altmetric Attention Score and how the score is calculated.

PDF (5 MB)

Get e-Alerts

Supporting Info (4)»Supporting Information Supporting Information

SUBJECTS:

Get e-Alerts

Abstract

A key metric to assess molecular docking remains ligand enrichment against challenging decoys. Whereas the directory of useful decoys (DUD) has been widely used, clear areas for optimization have emerged. Here we describe an improved benchmarking set that includes more diverse targets such as GPCRs and ion channels, totaling 102 proteins with 22886 clustered ligands drawn from ChEMBL, each with 50 property-matched decoys drawn from ZINC. To ensure chemotype diversity, we cluster each target’s ligands by their Bemis–Murcko atomic frameworks. We add net charge to the matched physicochemical properties and include only the most dissimilar decoys, by topology, from the ligands. An online automated tool (http://decoys.docking.org) generates these improved matched decoys for user-supplied ligands. We test this data set by docking all 102 targets, using the results to improve the balance between ligand desolvation and electrostatics in DOCK 3.6. The complete DUD-E benchmarking set is freely available at http://dude.docking.org.

Introduction

ARTICLE SECTIONS

Jump To

While molecular docking screens routinely leverage protein structure to discover new ligands, (1-4) quantitative assessment of their performance remains problematic. (5) Although prospective assessment of docking performance is irreplaceable, (6, 7) it is both time-consuming and expensive. Because a general correlation between docking scores and affinities is beyond current methods, (8, 9) the field relies on ligand enrichment in docking hit lists to evaluate retrospective performance. (10-14) “Enrichment” measures how known ligands rank versus a background of decoy molecules and so depends not only on the nature of the ligands but also on the background decoys. Thus to compare docking enrichments, a benchmarking set of ligands and decoys is needed.

The original Directory of Useful Decoys (DUD) was designed to meet this benchmarking need while controlling for decoy bias on enrichment. (15, 16) Given a random drug-like set of decoys, Verdonk et al. showed that targets which bind high molecular weight ligands naturally get higher enrichments due to correlation between larger molecules and better docking scores. (17) In contrast, actual ligand binding affinities correlate with molecular size only for very small molecules. (18) Unable to separate the true correlations of simple molecular properties that aid prospective ligand discovery from the artifical correlations that arise from biases, it is informative to ask what value molecular docking adds beyond these properties. To this end, DUD decoys are matched to the physical chemistry of ligands on a target-by-target basis: by the properties of molecular weight, calculated logP, number of rotatable bonds, and hydrogen bond donors and acceptors. To fulfill their role as negative controls, decoys should not actually bind, so DUD used 2-D similarity fingerprints to minimize the topological similarity between decoys and ligands. In short, DUD decoys were chosen to resemble ligands physically and so be challenging for docking but at the same time be topologically dissimilar to minimize the likelihood of actual binding.

Through intense use, (19-26) weaknesses in the original DUD set have appeared in both the ligands and decoys. Good and Oprea noted that a handful of chemotypes dominate many ligand sets, allowing high ranks for one scaffold to cause good overall enrichment. (27) One way to circumvent this problem is using chemotype retrieval metrics, (28) but another is to remove the “analogue bias” from the database by clustering on ligand scaffolds. After clustering the 40 targets, Good’s subset of DUD contains only 13 targets with over 15 ligands, indicating a need for more targets with more ligands. Another important goal is to increase target diversity, for example, by adding membrane domain proteins, none of which are represented in DUD.

As there were weaknesses in the DUD ligands, this was also true of the decoys. Several investigators (29-31) observed that despite property matching on logP, net formal charge is still imbalanced in DUD; 42% of all ligands are charged versus only 15% of decoys. Property matching of decoys to ligands could also be tightened by choosing decoys more embedded in ligand property space. (32, 33) Despite a 2-D chemical dissimilarity filter to prevent decoys from being active, some original DUD decoys still appear to bind, and these false decoys artificially reduce docking enrichment. (32) Addressing both false decoys and decoy property embedding, Vogel et al. released DEKOIS for the original 40 DUD targets. Gatica and Cavasotto generated ligand and decoy sets for 147 G protein-coupled receptors (GPCRs) while adding net charge to property matching. (34) Very recently, a python GUI application was announced to generate property-matched decoys. (35) By ignoring synthetic feasibility, Wallach and Lilien generate virtual decoy sets for the original DUD targets with tighter property-matching. (33) Instead of generating computational decoys, the MUV set selects decoys for 17 targets that were negative in public high-throughput screens. (36) Instead of generating decoys at all, REPROVIS-DB assembles ligand and database data from earlier successful virtual screens which are deemed reproducible. (37)

Here we describe a new version of DUD that addresses these liabilities and develops new functionality. By drawing on ChEMBL09, (38) each DUD-Enhanced (DUD-E) ligand has a measured affinity supported by a literature reference. Though ligands are now typically clustered by Bemis–Murcko atomic frameworks (39) to reduce chemotype bias, there are still on average 224 ligands per target. The target list is expanded from 40 to 102, favoring targets with many ligands and multiple (40) structures. The additions include several drug relevant membrane proteins: five GPCRs, two ion channels, and two cytochrome P450s. Meanwhile, false decoys are reduced by more stringent filtering of topological dissimilarity. Where possible, measured experimental decoys are included. Finally, we consider how DUD-E performs as a benchmark versus the original DUD and explore its use as a tool for evaluating and optimizing molecular docking.

Results

ARTICLE SECTIONS

Jump To

The ideal target for a benchmarking set would be well studied, with many measured ligand affinities and multiple, diverse cocrystal ligand structures. To this end, the enhanced DUD database (DUD-E) is largely based on the intersection of ChEMBL, (38) for ligand annotations and affinities, and the RCSB PDB, (40) for structures. As we sought targets to enlarge the set, the 40 original DUD targets were first priority, 38 of which we included. Platelet-derived growth factor receptor β was dropped, as it was a homology model. Estrogen receptor α (ESR1) is a single target in DUD-E, whereas it was split into agonists and antagonists previously. To enlarge the benchmarking set, we used three main criteria. First, we favored new target classes with pharmacological precedence. Second, we sought targets with many ligands and crystal structures, as they likely reflect a combination of target relevance and ease of study. Third, we preferred targets that could modestly enrich known ligands using fully automated docking, as these may be both easy to prepare and amenable to docking. Conversely, targets with mostly covalent ligands were deprioritized.

DUD-E targets are defined by their UniProt (41) gene prefix, with data from each species being combined into a single data set. While ChEMBL annotates ligands to a particular UniProt accession code, the ligand overlap between orthologous targets is surprisingly small. For example, among 1555 unique ligands with affinities below 1 μM for the human dopamine D3 receptor and 744 ligands for the rat orthologue, only 85 ligands are in both sets. These two orthologues share 97% trans-membrane sequence identity (79% overall), so this low overlap suggests to us that ChEMBL ligand annotations are sparse and do not typically reflect species specificity. Therefore, we pooled the data for all species, defining a DUD-E target as a UniProt gene prefix (such as DRD3), and not the full gene_species pair (such as DRD3_HUMAN or P35462).

The 102 targets span diverse protein categories, including 26 kinases, 15 proteases, 11 nuclear receptors, five GPCRs, two ion channels, two cytochrome P450s, 36 other enzymes, and five miscellaneous proteins (Figure 1). Altogether 66695 raw ligands, defined as those with annotated affinities better than 1 μM to their target, molecular weights less than 600 and fewer than 20 rotatable bonds were extracted from ChEMBL09 (or the AmpC β-lactamase literature) (Table 1). That is an average of 654 ligands per target with a minimum of 40 and a maximum of 3090. Though negative binding is rarely reported, we also found 9219 experimental decoys (i.e., no measurable affinity up to 30 μM), with a maximum of 1070 for cyclo-oxygenase-1 (PGH1).

Figure 1. DUD-E target classification. Number of the 102 targets that belong to eight broad protein categories.

Table 1. Characteristics of DUD-E

	total	ChEMBL	manual
no. targets	102	101	1

	total	average	minimum	maximum
no. raw ligands	66695	653.9	40	3090
no. clustered ligands	22886	224.4	40	592
no. experimental decoys	9219	90.4	1	1070
no. clustered ligands unique charge states	28377	278.2	46	1030
no. computational decoys	1411214	13835	2300	51500

With targets selected, we chose a single X-ray structure to represent each target in docking studies (Table 2, Supporting Information Table S1). To find the structure most amenable to docking, we used an automated docking campaign to screen 3690 PDB structures against their clustered ligands and property-matched decoys (see below). Preference was given to higher resolution, to higher automated enrichment, and to the human orthologue. We avoided mutant structures, unresolved active site loops, extraneous bound peptides, or structures too constrained for many of that target’s ligands. Where we had domain knowledge, the most representative structure was preferred, for example a DFG-in structure for kinases or an antagonist structure for estrogen receptor α (ESR1). For 57 out of 102 targets, a DOCK Blaster (42) prepared structure was used for DUD-E, directly from the automated tool chain. Another 45 targets required manual intervention, most due to simple errors in automated preparation (e.g., incomplete metal atom preparation, missing cofactors, or nonstandard amino acids). A select few needed expert intervention to arrive at modest enrichment, such as adding crystallographic waters, changing histidine protonation, flipping ambiguous side-chains such as asparagine, or increasing a local dipole moment on a specific residue (a technique we often use prospectively to improve polar complementarity (43, 44)). In five targets, we incorporated prior docking preparations used for prospective ligand discovery: adenosine A_2A receptor (AA2AR), (44) β₁ adrenergic receptor (ADRB1), AmpC β-lactamase (AMPC), C-X-C chemokine receptor type 4 (CXCR4), (3) and dopamine D₃ receptor (DRD3). (45)

Table 2. Overview of Representative Targets

target class	gene ID	description	total ligands	clustered ligands	experimental decoys	matched decoys	PDB	LogAUC (%)	ROC EF₁	AUC (%)
cytochrome P450	CP2C9	cytochrome P450 2C9	145	120	176	7450	1R9O	7	3	60
	CP3A4	cytochrome P450 3A4	302	170	267	11800	3NXU	7	2	63

GPCR	AA2AR	adenosine A2a receptor	3057	482	192	31550	3EML	28	22	83
	ADRB1	β-1 adrenergic receptor	648	247	69	15850	2VT4	19	11	76
	CXCR4	C-X-C chemokine receptor type 4	40	40	14	3406	3ODU	36	18	90

ion channel	GRIA2	glutamate receptor ionotropic AMPA 2	476	158	201	11845	3KGC	23	23	71
	GRIK1	glutamate receptor ionotropic kainate 1	136	101	235	6550	1VSO	35	27	86

kinase	AKT1	serine/threonine-protein kinase AKT	585	293	53	16450	3CQW	27	29	72
	MK10	c-Jun N-terminal kinase 3	199	104	23	6600	2ZDT	24	11	82
	MK14	MAP kinase p38 α	2205	578	73	35850	2QD9	17	10	74

miscellaneous	KIF11	kinesin-like protein 1	272	116	29	6850	3CJO	34	35	77
	XIAP	inhibitor of apoptosis protein 3	100	100	7	5150	3HL5	52	55	88

nuclear receptor	ESR1	estrogen receptor α	1297	383	136	20685	1SJ0	18	15	67
	MCR	mineralocorticoid receptor	201	94	2	5150	2AA2	–4	2	36
	THB	thyroid hormone receptor β-1	246	103	29	7450	1Q4X	36	38	79
	PPARD	peroxisome proliferator-activated receptor δ	699	240	79	12250	2ZNP	32	20	89

other enzymes	FNTA	protein farnesyltransferase type I α	1430	592	132	51500	3E37	16	7	76
	HDAC8	histone deacetylase 8	309	170	73	10450	3F07	29	24	80
	HIVINT	HIV type 1 integrase	167	100	268	6650	3NF7	8	2	64
	KITH	thymidine kinase	57	57	68	2850	2B8T	15	0	80
	PARP1	poly (ADP-ribose)polymerase-1	1031	508	12	30050	3L3M	25	21	79
	PUR2	GAR transformylase	50	50	12	2700	1NJS	51	50	92

protease	DPP4	dipeptidyl peptidase IV	1939	533	167	40950	2I78	41	41	87
	FA10	coagulation factor X	3090	537	176	28325	3KL6	39	36	87
	LKHA4	leukotriene A4 hydrolase	343	171	21	9450	3CHP	18	4	82
	MMP13	matrix metallo-proteinase 13	1632	572	26	37200	830C	12	5	71

To increase scaffold diversity and to make smaller, more manageable ligand sets, we clustered the raw ChEMBL ligands by their Bemis–Murcko atomic frameworks. (39) These atom-type based frameworks include ring systems of the molecule and connecting linkers, minus any side fragments. For example, the seventh largest Murcko cluster in kinesin-like protein 1 (KIF11) has seven ligands, all close analogues (Figure 2A). If at least 100 frameworks were present, then we included only the highest affinity ligand from each framework. If fewer were available, we raised the number of ligands selected from each framework until we obtained more than 100 molecules, trading diversity for quantity. Returning to kinesin-like protein 1, we extracted only 70 Murcko frameworks (Figure 2B). Out of 276 raw ligands, the five largest Murcko clusters contained 146 ligands (53%). Selecting the two or three highest affinity ligands from each framework results in 98 and 118 ligands, respectively, so we stopped at three ligands per framework. In the process we still managed to remove 158 lower affinity compounds from highly redundant clusters. In a few targets, more than 600 ligands remained even after clustering, so we reduced the affinity threshold below 1 μM in the sequence (300, 100, 30, 10, and 3 nM), until fewer than 600 frameworks were found. For example, in adenosine A_2A receptor, there are 3096 raw ligands resulting in 1099 frameworks at 1 μM, but we can reduce the number of frameworks to 483 using a 30 nM affinity threshold (Figure 2C).

Figure 2. Ligand clustering. (A) The seventh largest Murcko cluster of kinesin-like protein 1 (KIF11), showing both the scaffold (left) and all seven member ligands. (B) Number of ligands in each of the 70 KIF11 Bemis–Murcko atomic frameworks. We removed lower affinity compounds over-represented clusters (above the line), while retaining 100 ligands. (C) Number of adenosine A_2A receptor (AA2AR) Murcko clusters is plotted against affinity threshold. Fewer than 600 clusters are present using a 30 nM affinity threshold.

To examine the effect of clustering on docking enrichments, we docked the three targets with the highest and lowest fraction of clustered to raw ligands from those with enough ligands to pick one ligand per Murcko cluster. To measure docking performance we used LogAUC, an aggregate metric that gives early enrichment more weight. As described previously, (31) LogAUC is completely analogous to AUC but in the transformed space after you have zoomed in on early enrichment by taking the semilog of the x-axis. In tryptase β1 (TRYB1), the target with the highest clustered fraction, clustering substantially decreases the LogAUC by 6%, whereas in the other five targets clustering increases the LogAUC (Supporting Information Table S2). The mean absolute deviation over the six targets is 3.7% LogAUC, but in all cases the raw and clustered ROC curves have similar shapes (data not shown). Overall, we believe the clustered sets provide a better measure of docking performance with lower docking effort and will be used in the remainder of this work.

A key problem with the original DUD decoys was that they sometimes closely resembled the ligands, occasionally even being confirmed as binders. Enforcing 2-D topological dissimilarity between decoys and ligands should eliminate this problem in principle, but in practice critical ligand binding “warheads” often remain in the decoy set selected from ZINC, (46) e.g., amidine groups in factor Xa (FA10). By identifying these warheads in three targets (Figure 3A), we investigated how to eliminate false decoys. In the original DUD, CACTVS fingerprints were used to select decoys with Tanimoto coefficients (T_c) to ligands below 0.9, which is roughly similar to using Daylight fingerprints with T_c below 0.7. (15) In recent work, (31) we used Daylight fingerprints with a more restrictive T_c < 0.5. Using this filter on the enhanced DUD ligand sets, we still saw 39%, 53%, and 96% of possible warhead bearing molecules passing through in factor Xa (FA10), glycinamide ribonucleotide transformylase (PUR2), and thymidine kinase (KITH), respectively (Figure 3B). Using Daylight with T_c < 0.325, we reduced FA10 warheads below 1% but still saw 14% and 34% in PUR2 and KITH. Clearly different targets and even different ligands require different absolute thresholds. To circumvent this, we removed a percentage of the most similar decoys for each ligand, sorted by maximum T_c to any ligand. This allowed the effective absolute threshold to vary. Removing 50% of the decoys with Daylight was better in KITH, while removing 50% with ECFP4 was better in FA10 and PUR2. The final procedure of using ECFP4 fingerprints and removing 75% of the decoys, resulted in 0.2%, 0%, and 5.8% of warheads remaining, substantially reducing the number of false decoys. Having refined the decoy dissimilarity procedure on three targets where we could define a warhead, we then applied it to all generated decoys. To help ensure that the resulting decoys were, in fact, substantially different, topologically, from the ligands, we compared the two by a metric partially orthogonal to topology, asking how many decoy molecules shared the same scaffold as a ligand. Of the 805136 decoy scaffolds over all of DUD-E, only 692 (0.086%) were found among the 25503 ligand scaffolds, consistent with substantial topological differences among the two sets despite their close physical property matching.

Figure 3. Decoy generation. (A) Three key “warhead” groups from factor Xa (FA10), glycinamide ribonucleotide transformylase (PUR2), and thymidine kinase (KITH). (B) Fraction of warheads remaining is plotted against the dissimilarity method. The dissimilarity methods consist of a fingerprint (Daylight or ECFP4) and either a hard cutoff or a fraction of the most dissimilar decoys to be retained. (C) Property distributions of estrogen receptor α (ESR1) for both the 383 ligands (blue) and the 20685 property-matched decoys (red).

In addition to reducing false decoys, the DUD-E decoy generation procedure was extensively revised. Each decoy derived from a particular ligand, where decoy property ranges around the ligands properties adjusted to seven possible widths. This adapted to local chemical space around each ligand, allowing more closely matched decoys. Also, net charge was added to the property matching, as it is critical in electrostatics and desolvation. The improved property-matching can be seen in the property histograms for estrogen receptor α (ESR1) (Figure 3C) as well as the averages and standard deviations for all the targets (Supporting Information Table S3). Using ZINC (46) for the potential decoy pool made them purchasable, enabling experimental testing for actual binding to the target. As a result of this work, this enhanced decoy procedure has been fully automated and is available online to generate DUD-E style decoys for any user supplied list of input ligands at http://decoys.docking.org.

The original DUD paper (15) showed that a property-matched decoy set is more challenging for docking than a random collection of molecules. Therefore, we compared enrichments using property-matched decoys to those using a random drug-like background, which consisted of all ChEMBL12 ligands with affinities better than 10 μM. Switching from a drug-like background to DUD-E property-matched decoys does reduce average enrichment over the 102 targets, from 26.8% to 24.4% LogAUC (Supporting Information Table S4). Yet for three targets, the property-matched sets unexpectedly led to much better enrichment, by more than 15% LogAUC. In both glutamate receptor ionotropic kainate 1 (GRIK1) and purine nucleoside phosphorylase (PNPH), the ligands have low molecular weights (Supporting Information Table S3) and thus scored poorly against the generally larger ChEMBL12 molecules, just as Verdonk (17) suggests. In urokinase-type plasminogen activator (UROK), the top of the drug-like docking hit list is dominated by decoys with amidine “warheads”. Because these are likely binders, the increased property-matched enrichment resulted from fewer false decoys in that set. Indeed, the 2.4% LogAUC reduction that occurs upon switching to property-matched decoys arises from these two competing factors: property matching the decoys reduces enrichment, and reduction of false decoys increases enrichment.

Overall, enrichment as measured by average LogAUC is 1.5 fold higher in DUD-E compared to the original DUD. To understand this, we first isolated the change due to the revised decoy generation procedure. Using the original DUD ligands and target preparations, but switching from original decoys to these revised decoys substantially increased the average enrichment over the 37 directly comparable targets from 14.8% to 19.7% LogAUC (Table 3, Supporting Information Table S5). With the new adaptive property-matching procedure incorporating net charge, the revised decoys might have been expected to lower enrichment, but instead we saw an overall increase. Inspecting the docking hit lists, we observed a dramatic decrease in high scoring decoys that resemble ligands to a degree that they might actually bind. Indeed, all three targets with identifiable warheads that we used to tune the dissimilarity procedure showed large increases in enrichment: FA10 increases from 13% to 28% LogAUC, PUR2 from 40% to 62% LogAUC, and KITH from 1% to 32% LogAUC. If we now isolate the switch from original ligands and revised decoys to both DUD-E ligands and decoys, we see a moderate decrease in average enrichment from 19.7% to 16.4% LogAUC. We attribute this decrease to the larger, more diverse clustered ligand lists in DUD-E. Lastly, switching the target preparation, and the choice of the particular PDB structure used to represent a target, substantially increases enrichment from 16.4% to 22.8% LogAUC between DUD and DUD-E (Supporting Information Table S5). The overall effect with SEV ligand desolvation in DOCK 3.6 is to increase average enrichment from 14.8% LogAUC against DUD to 22.8% LogAUC against the DUD-E benchmark.

Table 3. Decomposition of Enrichment Changes between DUD and DUD-E

incremental change	all original	new style decoys	switch to new ligands	switch target preparation
decoys	DUD	DUD-E	DUD-E	DUD-E
ligands	DUD	DUD	DUD-E	DUD-E
receptor preparation	DUD	DUD	DUD	DUD-E
average LogAUC^a	14.8	19.7	16.4	22.8

Over the 37 common targets (target-by-target data in Supporting Information Table S5).

A central motivation for any benchmarking set is to test, at least retrospectively, new methods. We wanted to explore how our recent context-dependent ligand desolvation method (31) behaved against the DUD-E benchmark. We therefore used it to re-examine the utility of solvent-excluded volume (SEV) ligand desolvation versus using no desolvation term (None) or using the full transfer free energy from water to hexadecane (Full). In our initial study of these terms on the 40 original DUD targets, SEV improved upon None by just 0.7% average LogAUC. Conversely, over the 102 DUD-E targets, SEV substantially outperformed None by 3.8% LogAUC on average, with average LogAUC values of 20.6, 14.3, and 24.4% for None, Full, and SEV desolvation methods, respectively (Figure 4, Supporting Information Table S4). Despite these average trends, ROC curves on individual targets can vary significantly among the various methods (Figure 5). As in the original desolvation analysis, some targets are more amenable to full desolvation, such as catechol O-methyltransferase (COMT) and purine nucleoside phosphorylase (PNPH), while others are more amenable to no desolvation, such as factor X (FA10) and glycinamide ribonucleotide transformylase (PUR2). Against the DUD-E benchmark, SEV desolvation not only outperforms the other methods, but performs well in both types of targets. This suggests that over a more comprehensive set of targets, and what we argue is a better set of ligands and decoys, the advantage of the more physically correct SEV ligand desolvation treatment becomes more pronounced.

Figure 4. Retrospective enrichment comparing ligand desolvation and electrostatics methods. Docking results over DUD-E as measured by LogAUC. “None” has no ligand desolvation term, “SEV” uses solvent-excluded volume ligand desolvation, “Thin” employs a thin low-dielectric layer in the electrostatic calculations.

Figure 5. Representative ROC plots. ROC plots using no desolvation (None), solvent-excluded volume ligand desolvation (SEV), the thin low-dielectric layer (Thin), or a drug-like background that consists of all ChEMBL12 ligands with affinities better than 10 μM (Drug-like). The black dotted line represents the results expected from docking ligands randomly. LogAUC percentages are reported in the legend text.

Electrostatic interaction with the protein is a large term that opposes ligand desolvation, with their relative balance being critical for binding. Because we do not know the binding pose of putative ligands prior to docking, we need to approximate the region of low dielectric the ligand might occupy to precompute electrostatic grids. Previously, we used the negative image of the receptor (computed by SPHGEN) to construct this low dielectric region, but manual tweaking was often required. In the large open binding pocket of CXCR4, we observed that using a thin layer of low-dielectric around just the edge of the protein allowed ligands to interact with it while reducing the bulk dielectric perturbation at the center of its large binding pocket. (3) Here we explored using an automated thin dielectric layer strategy across the entire DUD-E set. Visually, these new automated thinner dielectric layers are more physically realistic, even in the rare case when they are effectively thicker than the previous layers (due to a water probe being able to penetrate that layer). With these thin low-dielectric layers (Thin), the average LogAUC over the 102 targets improved from 24.4% to 24.9% (Figure 4, Supporting Information Table S4). Six targets used manually prepared dielectric layers (AA2A2, ADRB1, AMPC, CDK2, CXCR4, and DRD3) and thus do not directly reflect the difference between automated dielectric layers. Excluding those six enlarges the average difference from 0.5% to 1.0% LogAUC. Admittedly, these are moderate differences, but they exemplify how DUD-E may be used to test new docking methods and hint that as we progress docking models, enrichment will improve.

Here we present three representative targets in greater detail to display a magnified view of DUD-E.

Mineralocorticoid Receptor (MCR)

MCR has the lowest enrichment in DUD-E. Across all 11 automatically docked structures, enrichment of DUD-E ligands to its decoys was negligible. Thus we selected the same PDB structure as the original DUD, 2AA2 at 1.95 Å resolution. While enrichment using the new DUD-E sets was worse than random at −4% LogAUC and 36% AUC (Table 2), using the original DUD ligands and decoys gave 45% LogAUC and 76% AUC. Despite poor enrichment in DUD-E, building and docking the crystal ligand from scratch, ignoring crystallographic information, resulted in good pose agreement (Figure 6A). Taken together, we can rationalize the enrichment differences, as 13 of 15 original ligands shared a polycyclic scaffold with the well-docked crystal ligand, while the 94 new ligands had much more scaffold diversity. Thus the reduced enrichment in DUD-E reflects increased chemotype diversity as a result of including more ligands and clustering them by Bemis–Murcko atomic frameworks. Of the four lowest enriching targets in DUD-E, three are nuclear hormone receptors, with glucocorticoid receptor (GCR) and androgen receptor (ANDR) joining MCR. These receptors all have hydrophobic pockets with flexible binding site residues such as methionine and leucine so that a single rigid receptor may be incapable of docking all of their ligands. Thus these targets may be good tests of flexible receptor docking methods.

Figure 6. Representative docking poses. The crystallographic ligand was rebuilt and docked from scratch. (A–F) The crystal pose (magenta) is compared to the resulting docked pose (green). In (C), more ligand conformations are generated and the redocked pose is also shown (tan). Key hydrogen bonds are shown by black dotted lines, and the partially transparent protein surface is colored by atom type.

Thyroid Hormone Receptor β1 (THB)

THB produced good enrichment when a structure with an open subpocket was selected. Enrichment for the 16 automatically docked structures varies significantly, ranging from 13% (1NQ0) to 37% LogAUC (1Q4X). The lower enriching structures have larger cavities near Arg320 (right side of Figure 6B), opening to solvent in 1NQ0; the higher enriching structures have larger cavities at the other end of the binding site near Met420 (left side), opening to solvent in 1Q4X. We selected the automated preparation of 1Q4X despite its modest 2.80 Å resolution because Thr273 is pushed away by the crystal ligand, making the left subpocket larger. Using SEV desolvation then yields enrichment statistics of 36% LogAUC, 79% AUC, and a receiver operating characteristic curve based enrichment factor at 1% (EF₁) of 38 (Table 2). The redocked crystal ligand has excellent pose agreement (Figure 6B).

Serine/Threonine-Protein Kinase AKT (AKT1)

AKT1 is a newly added kinase that demonstrates several considerations during PDB structure selection. Whereas 10 PDB structures were automatically docked, four got worse than random enrichment. All four correspond to structures of the Pleckstrin homology (PH) domain instead of the kinase domain. The structure with the best normal AUC, 3O96, corresponds to an allosteric site at the interface of the PH and kinase domains, not the traditional ATP binding pocket. While the best enriching structure by LogAUC, 3CQW at 2.00 Å, corresponds to the canonical site, its nonstandard phosphothreonine amino acid evades the automated protocol. Preparing that residue manually results in 27% LogAUC, 72% AUC, and 29 EF₁ (Table 2). Nevertheless, the redocked ligand (green) fails to generate the crystal ligand pose (magenta) (Figure 6C). The ligand, however, is quite small, with one central rotatable bond, and requires a specific rotation about that bond to fit in the binding site. Lowering the rmsd threshold for ligand conformation generation allows that rotation to be sampled, restoring the correct ligand binding pose (tan) (Figure 6C).

Discussion

ARTICLE SECTIONS

Jump To

Docking continues to be judged by hit rates in prospective studies, and by enrichment in retrospective recall studies, because it cannot now hope to calculate affinities or even monotonic rank order. Like protein structure prediction, docking thus remains an empirical, although we would argue also a pragmatic field. Its reliance on enrichment has driven the development of benchmarking sets, first explored by Rognan (11) and Jain, (12) recently investigated by Boeckler (32) and Cavasotto; (34) the most widely used and cited of these remains the Directory of Useful Decoys (DUD). (15) Despite its widespread adoption, DUD retains serious liabilities, including a lack of ligand diversity, lack of property-matching to net charge, and a substantial number of false decoys. The enhanced DUD (DUD-E) described here was developed to address these shortcomings and to expand the target list to be more reprentative of pharmacologically relevant space.

Balancing Ligands and Decoys for Enrichment

An important problem with DUD arose from the ligands and decoys originally chosen. The former sometimes over-represented in a few chemotypes, and the latter were sometimes not decoys but actually ligands. Moreover, the mapping of specific ligands to their matched decoys had been lost in the released set. In the 102 ligand and decoy sets that comprise DUD-E, ligand diversity in any given set is substantially increased, reducing the bias that can come from a single chemotype ranking well. With at least 40 ligands for every target and a preference to maximize chemotype diversity, DUD-E allows for more representative tests of docking screens. Correspondingly, property-matching decoys to each ligand individually, while more stringently removing false decoys (i.e., ligands), allows investigators to directly match specific ligands to their decoy molecules and reduces what had been artifactually low enrichment for some targets in DUD. Adding net charge as a property to match between ligands and decoys resolves a discrepancy between them in DUD, where the ligands had tended to be more charged, on average, than the decoys, which had the effect of skewing our evaluation of physical forces like desolvation.

The impact of these changes on docking performance is substantial and clarifying. In isolation from other effects, clustering the ligands for diversity reduces enrichment, as one might expect because high-performing, over-represented sets have been largely removed. Conversely, the new decoys increase enrichment compared to the DUD performance. At first this seemed counterintuitive, because one imagines that a better-balanced, more stringent decoy set will be a greater challenge for a docking program. However, this is more than balanced by the removal of what had been false decoys (ligands), which artifactually reduced enrichment in DUD because, as ligands, they had often ranked well but counted as decoys they diluted the annotated ligands. Finally, the new target preparations, carefully selected from a docking campaign to over 3500 structures, also increased enrichment. Overall, the increased enrichment in DUD-E should provide more sensitivity for benchmarking docking algorithms, giving it greater responsiveness to modifications that reduce enrichment as well as those that increase it.

Online Tools for Automated Generation of Further Ligand and Decoy Sets

DUD-E is built to be a better platform for refinement and extension of ligand and decoy sets. Targets are independent of one another, both in ligands and decoys, allowing target addition, deletion, or replacement. The protocol to generate decoys for DUD-E is made available online to generate decoys for any target given only a list of ligand structures, which enables extension of DUD-E to new targets of interest by individual investigators. The decoy server pulls directly from a purchasable subset of the ZINC database, inheriting its improvements and purchasing updates. (46) The final decoy selection from the applicable pool of decoys is random where possible, allowing the generation of multiple decoys sets to test overfitting to the canonical DUD-E decoys. Each decoy belongs to one and only one ligand, so if one wants to filter a ligand, then the corresponding decoys can be easily removed. For example, we provide raw ligand and decoy sets before clustering by Bemis–Murcko atomic frameworks. If a different clustering method was desired, which selected a different subset of the raw ligands, then the corresponding decoys could be retained (furthermore, we provide the python script used to generate clustered subsets from raw sets). We also include extra data that allows some design decisions to be altered, for instance, we include the marginal ligands which are active above our 1 μM cutoff.

Applications to Docking Optimization and Testing

DUD-E should provide a more robust benchmarking set for exploring new docking methods, so we were keen to test it against new methods that we had been investigating. When tested against the older DUD set, we had found that a new solvent-excluded volume (SEV) ligand desolvation method had had a disappointingly small effect on enrichment despite what was clearly a better physical model. However, when measured against the DUD-E benchmark, the differential performance between the old and new method increased substantially in the latter’s favor. Similarly, against the DUD-E benchmark, a more physically realistic dielectric layer, used to calculated the electrostatic interaction term from static Poission–Boltzmann maps, also led to improved enrichments that had been largely masked in the DUD set, owing to the problems described above.

Certain cautions merit airing. Most importantly, DUD-E is a large data set synthesized from several source databases, each of which is continuously evolving and improving. Thus individual errors are expected, though usually traceable to the source database at the time DUD-E was constructed. Although we only show docking results using DOCK 3.6 with solvent-excluded volume ligand desolvation, DUD-E was designed to be a general benchmarking set. Thus some arbitrary choices and simplifying assumptions were made in the effort to provide one canonical data set useful to compare docking algorithms. For instance, we assume a single PDB code can represent the target, but some targets are highly flexible or they contain both orthosteric and allosteric binding pockets. Fundamentally, DUD and DUD-E are designed to measure value-added screening performance of 3-D methods over simple 1-D molecular properties. Decoys that might bind are removed using 2-D ligand similarity, so DUD-E is inappropriate to test 2-D methods. Through its construction, ligands light up against DUD-E decoys using these 2-D similarity methods, which create an artificially favorable enrichment bias for them. A final caution is that to filter more false decoys in DUD-E, we keep only a quarter of the most highly dissimilar decoys. However, while we show that this increased dissimilarity removes false decoys, it could also contribute to artificial increases in docking enrichment.

Notwithstanding these caveats, DUD-E is substantially improved over the original DUD. It is a larger, more diverse data set with better matched decoys that resemble ligands less, correcting many flaws in its predecessor. Although we anticipate that it will be most widely used in the instantiation we describe here, it was developed with the idea that it could be flexibly extended and evolved; the tools to do so are even provided online (http://dude.docking.org). We hope that it and its descendants will provide a useful tool for docking evaluation in the community until such time as a more fundamental measurement of docking performance is possible.

Methods

ARTICLE SECTIONS

Jump To

ChEMBL and RCSB PDB Data Extraction

This enhanced DUD database has been constructed by combining ligand data from ChEMBL (38) and structural data from RCSB PDB (40) (Supporting Information Figure S1A). Ligands assigned to protein targets (ChEMBL confidence score ≥4) with affinities (IC₅₀, EC₅₀, K_i, K_d, and log variants thereof) of 1 μM or better were extracted from the ChEMBL09 database. (38) Similarly, we assigned experimental decoys as molecules with no measurable affinity at 30 μM or higher (greater than relation only). The remaining ligands with affinities above 1 μM, and decoys with no measurable affinity below 30 μM, are included for completeness and dubbed “marginal”. Via ChEMBL, ligands are associated with a particular target sequence by UniProt (41) accession code, and then mapped (47) from UniProt accession codes to protein data bank (PDB) structures (X-ray only) using http://www.uniprot.org/docs/pdbtosp.txt, obtained on February 23, 2011.

Target Selection Docking

Preliminary docking calculations were performed on each PDB structure that mapped to ChEMBL ligands and contained a single, unambiguous cocrystal ligand as prepared by DOCK Blaster. (42) Property-matched computational decoys were generated by the automated decoy generation procedure below, using Daylight fingerprints with a Tanimoto coefficient (T_c) threshold below 0.5. These decoys were docked and compared to their cognate ligands using DOCK 3.6 with solvent-excluded volume (SEV) ligand desolvation. (31) Balancing the parallel goals of diversity, drug relevance, many ligands and structures, and at least modest automated docking enrichment, we selected 119 tentative targets for the new DUD. This list was reduced to the final 102 targets by factors such as ligand and PDB duplication between targets (e.g., FNTB duplicates FNTA), low resolution structures (RAF1), sterically constrained binding sites (NR1H2, THA), or over-representation (MK08, MTOR).

Target Preparation

For each target, we assembled all UniProt accession codes (species) with any raw ChEMBL compounds (ligands, decoys, marginal ligands, or marginal decoys). For only those accession codes, structures were extracted using the ChEMBL to PDB mapping, except P07700 was manually added to ADRB1 to include six more rare structures for that GPCR. This procedure neglects those PDB structures that belong to an accession code having no ChEMBL compounds. For example, 1KIM is the PDB structure of thymidine kinase (KITH) in the original DUD. This KITH structure is from herpes virus (UniProt P03176), an accession code with no raw compounds extracted from ChEMBL, and is thus not included in the ChEMBL/PDB intersection used to construct the new DUD. Still, 5025 PDB codes were sent to an updated DOCK Blaster pipeline for automated docking preparation (Supporting Information Figure S1D). In some cases, an unambiguous ligand could not be found to indicate the binding site, but we were able to assign 565 additional ligands by manually inspecting over 1300 structures. Ultimately, 3692 structures completed input grid preparation, and all but two finished docking and enrichment analysis. Clustered ligands sets were docked to property-matched decoys (both described below) using ECFP4 fingerprints and removing the most similar 75% of queried decoys. DOCK 3.6 was run using SEV ligand desolvation (as below). For each target, enrichment, resolution, and organism were collected and sorted by enrichment in pdb_analyze.txt, available online at http://dude.docking.org. Crude notes on the selection process are recorded in pdb_selection.txt, and the picked structure is listed in pdb_blessed.txt. AA2AR and DRD3 docking preparations were provided by Jens Carlson, (44, 45) CXCR4 partially by Dahlia Weiss, (3) ADRB1 by Peter Kolb (personal communication), and AMPC by Sarah Barelier, Oliv Eidam, and Inbar Fish (unpublished results).

Ligand Preparation

To prepare ligand sets for each target, ChEMBL affinities and log variants were first normalized to nM units (Supporting Information Figure S1B). Salts were removed, charges were normalized, and properties were calculated using Molinspiration’s mib package (www.molinspiration.com). Ligands with 600 Da or higher molecular weight or with 20 or more rotatable bonds were removed. Smiles were put in canonical form using OpenEye’s OEChem software. (48) Ligand sets from each species were combined, sorted by ascending normalized affinity, and then made unique based on canonical smiles. The same procedure was used to collate the experimental decoys, marginal ligands, and marginal decoys. For AmpC β-lactamase (AMPC), an original DUD target, the ChEMBL09 ligands are covalent in nature. To identify noncovalent ligands, we manually compiled ligands (6, 43, 49, 50) with affinities below 5 mM and experimental decoys (43, 51) from the literature.

Ligand Clustering

To reduce the sometimes large number of ChEMBL ligands down to a manageable size while also increasing scaffold diversity as suggested by Good and Oprea, (27) we clustered the ligands by their Bemis–Murcko atomic frameworks, (39) as generated by Molinspiration’s mib. If there were 100 or more frameworks, we chose only the highest affinity ligand from each. If there were fewer than 100 Murcko frameworks, we increased the number of highest affinity ligands taken from each until we achieved at least 100 ligands (or until all ligands were included). Conversely, if there were more than 600 Murcko frameworks, then we decreased the ligand affinity threshold in the sequence [1 μM, 300 nM, 100 nM, 30 nM, 10 nM, 3 nM] until fewer than 600 frameworks were present, where we then took the highest affinity ligand from each framework. While clustered ligand sets are the default, the full unclustered ligand sets and corresponding decoys are available. The script (subset_decoys.py) used to select the clustered subset given the ligand ids is provided with the full ligand set to enable other clustering algorithms or filtering methods to be substituted.

Automated Decoy Generation

As in the original DUD, we property-matched decoys to ligands using molecular weight, estimated water–octanol partition coefficient (miLogP), rotatable bonds, hydrogen bond acceptors, and hydrogen bond donors, plus we added net charge. We generated all ligand protonation states in pH range 6–8 using Schrödinger’s Epik with arguments “-ph 7.0 -pht 1.0 -tp 0.20” (Supporting Information Figure S1C). Molecular properties were then computed using Molinspiration’s mib. Over all the protonated forms of a given ligand, we kept only those with a unique set of the six physicochemical properties. For each of these unique property sets, we aimed to generate 50 matched decoys. For example, a single input ligand predicted to have two alternate charges would get 50 decoys property-matched to each charge. To accomplish this, a pool of decoys was selected from ZINC (46) using a dynamic protocol that adapted to local chemical space by narrowing or widening windows in seven steps around the six properties. The goal was to return 3000–9000 potential decoys that matched the decoy’s reference protonation state (predicted most prevalent form at pH 7.05). In the final decoy procedure, ECFP4 fingerprints were generated by Scitegic’s Pipeline Pilot for ligands and potential decoys. The decoys were sorted by their maximum T_c to any ligand, and the most dissimilar 25% were retained through this dissimilarity filter. We then remove duplicate decoys from the ligand set by sorting decoys from least to most duplicated and assigned each decoy to the protonated ligand which has the least number of decoys already assigned. This ensures unique decoys were spread across the ligands as evenly as possible. Finally, if available, 50 decoys were picked randomly from this deduplicated list.

Original DUD Comparison

For the original DUD comparison, we downloaded ligands and decoys from dud.docking.org and prepared docking flexibases with our modern ZINC toolchain. (46) The original DUD target preparations were copies of the original, modified to perform SEV desolvation calculations as described previously. (31) We also generated DUD-E style automated decoys and flexibases for the original DUD ligands. The analysis was performed on the 37 directly comparable targets, excluding the original targets PDGFrb, ER_agonist, and ER_antagonist.

Docking Calculations

Except as noted, docking calculations were performed with DOCK 3.6 and solvent-excluded volume (SEV) ligand desolvation as described previously. (31) Ligand conformations were generated by OpenEye’s Omega. (52) For sampling, the minimum number of graph matching nodes was changed to 3, and ligand overlap was changed to 0.1. Ligands were limited to between 5 and 100 heavy atoms. The timeout for an individual ligand hierarchy was 180 s. We performed 200 steps of simplex minimization, with initial translations of 0.2 Å and initial rotations of 5°. The thin dielectric layer Delphi spheres were created by walking out each DMS (http://www.cgl.ucsf.edu/Overview/ftp/dms.zip) surface normal by 1.8 Å and placing a sphere. This thin sphere layer is then used as input to makespheres1.pl in place of the usual SPHGEN spheres. The random background calculations were performed using SEV desolvation by seeding the DUD-E ligands into the entire ChEMBL12_10 subset of ZINC, which includes 273375 ligands with annotated affinities below 10 μM.

Docking Metrics

The area under the curve (AUC) of the receiver operating characteristic (ROC) is one common metric to measure docking performance. However, ROC plots often use a semilog transformation of the x-axis to zoom in on early changes. As described previously, (31) LogAUC is completely analogous to AUC in this transformed space, measuring the percentage of the unit area under the curve. Formally, we use the adjusted LogAUC_0.001 here, which spans three decades of log space and subtracts the LogAUC of the random curve (14.462%) so that random enrichment is 0%. We typically refer to the adjusted LogAUC_0.001 as either adjusted LogAUC or simply LogAUC. The ROC-based enrichment factor at 1% (EF₁) is the percent of ligands found when 1% of the decoys have been found and is preferred over traditional enrichment factors. (53)

Supporting Information

ARTICLE SECTIONS

Jump To

Figure showing DUD-E workflows, while tables provide detailed target-by-target data and tab delimited text files provide the raw data. This material is available free of charge via the Internet at http://pubs.acs.org.

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 Authors
- John. J. Irwin - Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, California 94158-2330, United States; Email: [email protected] [email protected]
- Brian K. Shoichet - Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, California 94158-2330, United States; Email: [email protected] [email protected]
Authors
- Michael M. Mysinger - Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, California 94158-2330, United States
- Michael Carchia - Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, California 94158-2330, United States
Notes
The authors declare no competing financial interest.

Acknowledgment

ARTICLE SECTIONS

Jump To

Supported by NIH grant GM71896 (to J.J.I. and B.K.S.). We thank Andrew Good for discussions that initiated DUD-E. We thank Teague Sterling for website development and Sunil Koovakkat for DOCK bugfixes. We are grateful to the commercial software vendors who support ZINC and the decoy generation toolchain: Molinspiration (Bratislava, Slovakia) for mib, OpenEye Scientific Software (Santa Fe, NM) for OEChem, Omega, and QuacPac, Molecular Networks (Erlangen, Germany) for Corina, Accelrys (San Diego, CA) for Pipeline Pilot, and ChemAxon (Budapest, Hungary) for cxcalc. We thank Oliv Eidam, Matthew Merski, and Nir London for reading this manuscript.

Abbreviations Used

ARTICLE SECTIONS

Jump To

DUD	Directory of Useful Decoys
DUD-E	Directory of Useful Decoys—Enhanced
EF₁	enrichment factor at 1% of ROC curve
PH	pleckstrin homology
ROC	receiver operating characteristic
SEV	solvent-excluded volume
T_c	Tanimoto coefficient

References

ARTICLE SECTIONS

Jump To

This article references 53 other publications.

1
Kitchen, D. B.; Decornez, H.; Furr, J. R.; Bajorath, J. Docking and scoring in virtual screening for drug discovery: methods and applications Nature Rev. Drug Discovery 2004, 3, 935– 949
Google Scholar
1
Docking and scoring in virtual screening for drug discovery: methods and applications
Kitchen, Douglas B.; Decornez, Helene; Furr, John R.; Bajorath, Juergen
Nature Reviews Drug Discovery (2004), 3 (11), 935-949CODEN: NRDDAG; ISSN:1474-1776. (Nature Publishing Group)
A review. Computational approaches that 'dock' small mols. into the structures of macromol. targets and 'score' their potential complementarity to binding sites are widely used in hit identification and lead optimization. Indeed, there are now a no. of drugs whose development was heavily influenced by or based on structure-based design and screening strategies, such as HIV protease inhibitors. Nevertheless, there remain significant challenges in the application of these approaches, in particular in relation to current scoring schemes. Here, we review key concepts and specific features of small-mol.-protein docking methods, highlight selected applications and discuss recent advances that aim to address the acknowledged limitations of established approaches.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXptFemtrg%253D&md5=875a2b37a4299181509a1922b11dbd2f
2
Kolb, P.; Rosenbaum, D. M.; Irwin, J. J.; Fung, J. J.; Kobilka, B. K.; Shoichet, B. K. Structure-based discovery of beta(2)-adrenergic receptor ligands Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 6843– 6848
Google Scholar
2
Structure-based discovery of β2-adrenergic receptor ligands
Kolb, Peter; Rosenbaum, Daniel M.; Irwin, John J.; Fung, Juan Jose; Kobilka, Brian K.; Shoichet, Brian K.
Proceedings of the National Academy of Sciences of the United States of America (2009), 106 (16), 6843-6848CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
Aminergic G protein-coupled receptors (GPCRs) have been a major focus of pharmaceutical research for many years. Due partly to the lack of reliable receptor structures, drug discovery efforts have been largely ligand-based. The recently detd. X-ray structure of the β2-adrenergic receptor offers an opportunity to investigate the advantages and limitations inherent in a structure-based approach to ligand discovery against this and related GPCR targets. Approx. 1 million com. available, "lead-like" mols. were docked against the β2-adrenergic receptor structure. On testing of 25 high-ranking mols., 6 were active with binding affinities <4 μM, with the best mol. binding with a Ki of 9 nM (95% confidence interval 7-10 nM). Five of these mols. were inverse agonists. The high hit rate, the high affinity of the most potent mol., the discovery of unprecedented chemotypes among the new inhibitors, and the apparent bias toward inverse agonists among the docking hits, have implications for structure-based approaches against GPCRs that recognize small org. mols.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXlsV2qsro%253D&md5=4d1a4cb2aa3925aa4c99c6b0496417a7
3
Mysinger, M. M.; Weiss, D. R.; Ziarek, J. J.; Gravel, S.; Doak, A. K.; Karpiak, J.; Heveker, N.; Shoichet, B. K.; Volkman, B. F. Structure-based ligand discovery for the protein–protein interface of chemokine receptor CXCR4 Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 5517– 5522
Google Scholar
There is no corresponding record for this reference.
4
Gruneberg, S.; Stubbs, M. T.; Klebe, G. Successful virtual screening for novel inhibitors of human carbonic anhydrase: strategy and experimental confirmation J. Med. Chem. 2002, 45, 3588– 3602
Google Scholar
There is no corresponding record for this reference.
5
Jain, A. N.; Nicholls, A. Recommendations for evaluation of computational methods J. Comput.-Aided Mol. Des. 2008, 22, 133– 139
Google Scholar
5
Recommendations for evaluation of computational methods
Jain, Ajay N.; Nicholls, Anthony
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 133-139CODEN: JCADEQ; ISSN:0920-654X. (Springer)
A review. The field of computational chem., particularly as applied to drug design, has become increasingly important in terms of the practical application of predictive modeling to pharmaceutical research and development. Tools for exploiting protein structures or sets of ligands known to bind particular targets can be used for binding-mode prediction, virtual screening, and prediction of activity. A serious weakness within the field is a lack of stds. with respect to quant. evaluation of methods, data set prepn., and data set sharing. Our goal should be to report new methods or comparative evaluations of methods in a manner that supports decision making for practical applications. Here we propose a modest beginning, with recommendations for requirements on statistical reporting, requirements for data sharing, and best practices for benchmark prepn. and usage.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsb0%253D&md5=9576e150079dcbf6e4bfea5070082016
6
Babaoglu, K.; Simeonov, A.; Irwin, J. J.; Nelson, M. E.; Feng, B.; Thomas, C. J.; Cancian, L.; Costi, M. P.; Maltby, D. A.; Jadhav, A.; Inglese, J.; Austin, C. P.; Shoichet, B. K. Comprehensive mechanistic analysis of hits from high-throughput and docking screens against beta-lactamase J. Med. Chem. 2008, 51, 2502– 2511
Google Scholar
6
Comprehensive Mechanistic Analysis of Hits from High-Throughput and Docking Screens against β-Lactamase
Babaoglu, Kerim; Simeonov, Anton; Irwin, John J.; Nelson, Michael E.; Feng, Brian; Thomas, Craig J.; Cancian, Laura; Costi, M. Paola; Maltby, David A.; Jadhav, Ajit; Inglese, James; Austin, Christopher P.; Shoichet, Brian K.
Journal of Medicinal Chemistry (2008), 51 (8), 2502-2511CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
High-throughput screening (HTS) is widely used in drug discovery. Esp. for screens of unbiased libraries, false positives can dominate "hit lists"; their origins are much debated. Here we det. the mechanism of every active hit from a screen of 70,563 unbiased mols. against β-lactamase using quant. HTS (qHTS). Of the 1274 initial inhibitors, 95% were detergent-sensitive and were classified as aggregators. Among the 70 remaining were 25 potent, covalent-acting β-lactams. Mass spectra, counter-screens, and crystallog. identified 12 as promiscuous covalent inhibitors. The remaining 33 were either aggregators or irreproducible. No specific reversible inhibitors were found. We turned to mol. docking to prioritize mols. from the same library for testing at higher concns. Of 16 tested, 2 were modest inhibitors. Subsequent X-ray structures corresponded to the docking prediction. Analog synthesis improved affinity to 8 μM. These results suggest that it may be the phys. behavior of org. mols., not their reactivity, that accounts for most screening artifacts. Structure-based methods may prioritize weak-but-novel chemotypes in unbiased library screens.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjtFSksb0%253D&md5=7736429aeafc6b17a5d0182e24df0dd4
7
Ferreira, R. S.; Simeonov, A.; Jadhav, A.; Eidam, O.; Mott, B. T.; Keiser, M. J.; McKerrow, J. H.; Maloney, D. J.; Irwin, J. J.; Shoichet, B. K. Complementarity between a docking and a high-throughput screen in discovering new cruzain inhibitors J. Med. Chem. 2010, 53, 4891– 4905
Google Scholar
7
Complementarity Between a Docking and a High-Throughput Screen in Discovering New Cruzain Inhibitors
Ferreira, Rafaela S.; Simeonov, Anton; Jadhav, Ajit; Eidam, Oliv; Mott, Bryan T.; Keiser, Michael J.; McKerrow, James H.; Maloney, David J.; Irwin, John J.; Shoichet, Brian K.
Journal of Medicinal Chemistry (2010), 53 (13), 4891-4905CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Virtual and high-throughput screens (HTS) should have complementary strengths and weaknesses, but studies that prospectively and comprehensively compare them are rare. We undertook a parallel docking and HTS screen of 197861 compds. against cruzain, a thiol protease target for Chagas disease, looking for reversible, competitive inhibitors. On workup, 99% of the hits were eliminated as false positives, yielding 146 well-behaved, competitive ligands. These fell into five chemotypes: two were prioritized by scoring among the top 0.1% of the docking-ranked library, two were prioritized by behavior in the HTS and by clustering, and one chemotype was prioritized by both approaches. Detn. of an inhibitor/cruzain crystal structure and comparison of the high-scoring docking hits to expt. illuminated the origins of docking false-negatives and false-positives. Prioritizing mols. that are both predicted by docking and are HTS-active yields well-behaved mols., relatively unobscured by the false-positives to which both techniques are individually prone.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXntlWrurk%253D&md5=0661663ff0802bf981a1997eeca9fa05
8
Gohlke, H.; Klebe, G. Approaches to the description and prediction of the binding affinity of small-molecule ligands to macromolecular receptors Angew. Chem., Int. Ed. Engl. 2002, 41, 2644– 2676
Google Scholar
8
Approaches to the description and prediction of the binding affinity of small-molecule ligands to macromolecular receptors
Gohlke, Holger; Klebe, Gerhard
Angewandte Chemie, International Edition (2002), 41 (15), 2644-2676CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH)
A review. The influence of a xenobiotic compd. on an organism is usually summarized by the expression biol. activity. If a controlled, therapeutically relevant, and regulatory action is obsd. the compd. has potential as a drug, otherwise its toxicity on the biol. system is of interest. However, what do we understand by the biol. activity. In principle, the overall effect on an organism has to be considered. However, because of the complexity of the interrelated processes involved, as a simplification primarily the "main action" on the organism is taken into consideration. On the mol. level, biol. activity corresponds to the binding of a (lowmol. wt.) compd. to a macromol. receptor, usually a protein. Enzymic reactions or signal-transduction cascades are thereby influenced with respect to their function for the organism. We regard this binding as a process under equil. conditions; thus, binding can be described as an assocn. or dissocn. process. Accordingly, biol. activity is expressed as the affinity of both partners for each other, as a thermodn. equil. quantity. How well do we understand these terms and how well are they theor. predictable today. The holy grail of rational drug design is the prediction of the biol. activity of a compd. The processes involving ligand binding are extremely complicated, both ligand and protein are flexible mols., and the energy inventory between the bound and unbound states must be considered in aq. soln. How sophisticated and reliable are our exptl. approaches to obtaining the necessary insight. The present review summarizes our current understanding of the binding affinity of a small-mol. ligand to a protein. Both theor. and empirical approaches for predicting binding affinity, starting from the three-dimensional structure of a protein-ligand complex, will be described and compared. Exptl. methods, primarily microcalorimetry, will be discussed. As a perspective, our own knowledge-based approach towards affinity prediction and exptl. data on factorizing binding contributions to protein-ligand binding will be presented.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38Xmt1entbw%253D&md5=d2e5e2490fe8909e5899a1a89630223f
9
Enyedy, I. J.; Egan, W. J. Can we use docking and scoring for hit-to-lead optimization? J. Comput.-Aided Mol. Des. 2008, 22, 161– 168
Google Scholar
9
Can we use docking and scoring for hit-to-lead optimization?
Enyedy, Istvan J.; Egan, William J.
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 161-168CODEN: JCADEQ; ISSN:0920-654X. (Springer)
Docking and scoring is currently one of the tools used for hit finding and hit-to-lead optimization when structural information about the target is known. Docking scores have been found useful for optimizing ligand binding to reproduce exptl. obsd. binding modes. The question is, can docking and scoring be used reliably for hit-to-lead optimization To illustrate the challenges of scoring for hit-to-lead optimization, the relationship of docking scores with exptl. detd. IC50 values measured inhouse were tested. The influences of the particular target, crystal structure, and the precision of the scoring function on the ability to differentiate between actives and inactives were analyzed by calcg. the area under the curve of receiver operator characteristic curves for docking scores. It was found that for the test sets considered, MW and sometimes ClogP were as useful as GlideScores and no significant difference was obsd. between SP and XP scores for differentiating between actives and inactives. Interpretation by an expert is still required to successfully utilize docking and scoring in hit-to-lead optimization.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsbk%253D&md5=24adfd0e8bb71c0e7b32a9208e7ab4d1
10
Stahl, M.; Rarey, M. Detailed analysis of scoring functions for virtual screening J. Med. Chem. 2001, 44, 1035– 1042
Google Scholar
10
Detailed Analysis of Scoring Functions for Virtual Screening
Stahl, Martin; Rarey, Matthias
Journal of Medicinal Chemistry (2001), 44 (7), 1035-1042CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
The authors present a comprehensive study of the performance of fast scoring functions for library docking using the program FlexX as the docking engine. Four scoring functions, among them two recently developed knowledge-based potentials, are evaluated on seven target proteins whose binding sites represent a wide range of size, form, and polarity. The results of these calcns. give valuable insight into strengths and weaknesses of current scoring functions. Furthermore, it is shown that a well-chosen combination of two of the tested scoring functions leads to a new, robust scoring scheme with superior performance in virtual screening.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXhsFKitro%253D&md5=9fe0ea156ae846868e710115e1d71500
11
Bissantz, C.; Folkers, G.; Rognan, D. Protein-based virtual screening of chemical databases. 1. Evaluation of different docking/scoring combinations J. Med. Chem. 2000, 43, 4759– 4767
Google Scholar
11
Protein-Based Virtual Screening of Chemical Databases. 1. Evaluation of Different Docking/Scoring Combinations
Bissantz, Caterina; Folkers, Gerd; Rognan, Didier
Journal of Medicinal Chemistry (2000), 43 (25), 4759-4767CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Three different database docking programs (Dock, FlexX, Gold) have been used in combination with seven scoring functions (Chemscore, Dock, FlexX, Fresno, Gold, Pmf, Score) to assess the accuracy of virtual screening methods against two protein targets (thymidine kinase, estrogen receptor) of known three-dimensional structure. For both targets, it was generally possible to discriminate about 7 out of 10 true hits from a random database of 990 ligands. The use of consensus lists common to two or three scoring functions clearly enhances hit rates among the top 5% scorers from 10% (single scoring) to 25-40% (double scoring) and up to 65-70% (triple scoring). However, in all tested cases, no clear relationships could be found between docking and ranking accuracies. Moreover, predicting the abs. binding free energy of true hits was not possible whatever docking accuracy was achieved and scoring function used. As the best docking/consensus scoring combination varies with the selected target and the physicochem. of target-ligand interactions, we propose a two-step protocol for screening large databases: (i) screening of a reduced dataset contg. a few known ligands for deriving the optimal docking/consensus scoring scheme, (ii) applying the latter parameters to the screening of the entire database.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXotFymurY%253D&md5=ee74ac9a99e55759c2df27fd1db58010
12
Pham, T. A.; Jain, A. N. Parameter estimation for scoring protein–ligand interactions using negative training data J. Med. Chem. 2006, 49, 5856– 5868
Google Scholar
12
Parameter Estimation for Scoring Protein-Ligand Interactions Using Negative Training Data
Pham, Tuan A.; Jain, Ajay N.
Journal of Medicinal Chemistry (2006), 49 (20), 5856-5868CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Surflex-Dock employs an empirically derived scoring function to rank putative protein-ligand interactions by flexible docking of small mols. to proteins of known structure. The scoring function employed by Surflex was developed purely on the basis of pos. data, comprising noncovalent protein-ligand complexes with known binding affinities. Consequently, scoring function terms for improper interactions received little wt. in parameter estn., and an ad hoc scheme for avoiding protein-ligand interpenetration was adopted. We present a generalized method for incorporating synthetically generated neg. training data, which allows for rigorous estn. of all scoring function parameters. Geometric docking accuracy remained excellent under the new parametrization. In addn., a test of screening utility covering a diverse set of 29 proteins and corresponding ligand sets showed improved performance. Maximal enrichment of true ligands over non-ligands exceeded 20-fold in over 80% of cases, with enrichment of greater than 100-fold in over 50% of cases.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXlsF2ls70%253D&md5=2296d0e658abfda21e3cc50d2a248e44
13
Kellenberger, E.; Rodrigo, J.; Muller, P.; Rognan, D. Comparative evaluation of eight docking tools for docking and virtual screening accuracy Proteins 2004, 57, 225– 242
Google Scholar
13
Comparative evaluation of eight docking tools for docking and virtual screening accuracy
Kellenberger, Esther; Rodrigo, Jordi; Muller, Pascal; Rognan, Didier
Proteins: Structure, Function, and Bioinformatics (2004), 57 (2), 225-242CODEN: PSFBAF ISSN:. (Wiley-Liss, Inc.)
Eight docking programs (DOCK, FLEXX, FRED, GLIDE, GOLD, SLIDE, SURFLEX, and QXP) that can be used for either single-ligand docking or database screening have been compared for their propensity to recover the x-ray pose of 100 small-mol.-wt. ligands, and for their capacity to discriminate known inhibitors of an enzyme (thymidine kinase) from randomly chosen "drug-like" mols. Interestingly, both properties are found to be correlated, since the tools showing the best docking accuracy (GLIDE, GOLD, and SURFLEX) are also the most successful in ranking known inhibitors in a virtual screening expt. Moreover, the current study pinpoints some physicochem. descriptors of either the ligand or its cognate protein-binding site that generally lead to docking/scoring inaccuracies.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXnvFaku7k%253D&md5=75a11a8f0603c819e4f4c8013b948686
14
Ferrara, P.; Gohlke, H.; Price, D. J.; Klebe, G.; Brooks, C. L., III. Assessing scoring functions for protein–ligand interactions J. Med. Chem. 2004, 47, 3032– 3047
Google Scholar
14
Assessing Scoring Functions for Protein-Ligand Interactions
Ferrara, Philippe; Gohlke, Holger; Price, Daniel J.; Klebe, Gerhard; Brooks, Charles L., III
Journal of Medicinal Chemistry (2004), 47 (12), 3032-3047CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
An assessment of nine scoring functions commonly applied in docking using a set of 189 protein-ligand complexes is presented. The scoring functions include the CHARMm potential, the scoring function DrugScore, the scoring function used in AutoDock, the three scoring functions implemented in DOCK, as well as three scoring functions implemented in the CScore module in SYBYL (PMF, Gold, ChemScore). The authors evaluated the abilities of these scoring functions to recognize near-native configurations among a set of decoys and to rank binding affinities. Binding site decoys were generated by mol. dynamics with restraints. To investigate whether the scoring functions can also be applied for binding site detection, decoys on the protein surface were generated. The influence of the assignment of protonation states was probed by either assigning "std." protonation states to binding site residues or adjusting protonation states according to exptl. evidence. The role of solvation models in conjunction with CHARMm was explored in detail. These include a distance-dependent dielec. function, a generalized Born model, and the Poisson equation. The authors evaluated the effect of using a rigid receptor on the outcome of docking by generating all-pairs decoys ("cross-decoys") for six trypsin and seven HIV-1 protease complexes. The scoring functions perform well to discriminate near-native from misdocked conformations, with CHARMm, DOCK-energy, DrugScore, ChemScore, and AutoDock yielding recognition rates of around 80%. Significant degrdn. in performance is obsd. in going from decoy to cross-decoy recognition for CHARMm in the case of HIV-1 protease, whereas DrugScore and ChemScore, as well as CHARMm in the case of trypsin, show only small deterioration. In contrast, the prediction of binding affinities remains problematic for all of the scoring functions. ChemScore gives the highest correlation value with R2 = 0.51 for the set of 189 complexes and R2 = 0.43 for the set of 116 complexes that does not contain any of the complexes used to calibrate this scoring function. Neither a more accurate treatment of solvation nor a more sophisticated charge model for zinc improves the quality of the results. Improved modeling of the protonation states, however, leads to a better prediction of binding affinities in the case of the generalized Born and the Poisson continuum models used in conjunction with the CHARMm force field. The method can be used for drug discovery.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXjs1Sjsrk%253D&md5=a9f1e74194d3949aefadefb3266372d2
15
Huang, N.; Shoichet, B. K.; Irwin, J. J. Benchmarking sets for molecular docking J. Med. Chem. 2006, 49, 6789– 6801
Google Scholar
15
Benchmarking Sets for Molecular Docking
Huang, Niu; Shoichet, Brian K.; Irwin, John J.
Journal of Medicinal Chemistry (2006), 49 (23), 6789-6801CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Ligand enrichment among top-ranking hits is a key metric of mol. docking. To avoid bias, decoys should resemble ligands phys., so that enrichment is not simply a sepn. of gross features, yet be chem. distinct from them, so that they are unlikely to be binders. We have assembled a directory of useful decoys (DUD), with 2950 ligands for 40 different targets. Every ligand has 36 decoy mols. that are phys. similar but topol. distinct, leading to a database of 98 266 compds. For most targets, enrichment was at least half a log better with uncorrected databases such as the MDDR than with DUD, evidence of bias in the former. These calcns. also allowed 40×40 cross-docking, where the enrichments of each ligand set could be compared for all 40 targets, enabling a specificity metric for the docking screens. DUD is freely available online as a benchmarking set for docking at http://blaster.docking.org/dud/.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28XhtFehurzI&md5=43af46968c2a6d66334922bf0caedbe1
16
Christofferson, A. J.; Huang, N. How to benchmark methods for structure-based virtual screening of large compound libraries. In Computational Drug Discovery and Design (Methods in Molecular Biology); 2011/12/21 ed.; Baron, R., Ed.; Springer Protocols: New York, 2012; Vol. 819, Chapter 13, pp 187– 195.
Google Scholar
There is no corresponding record for this reference.
17
Verdonk, M. L.; Berdini, V.; Hartshorn, M. J.; Mooij, W. T.; Murray, C. W.; Taylor, R. D.; Watson, P. Virtual screening using protein–ligand docking: avoiding artificial enrichment J. Chem. Inf. Comput. Sci. 2004, 44, 793– 806
Google Scholar
17
Virtual screening using protein-ligand docking: Avoiding artificial enrichment
Verdonk, Marcel L.; Berdini, Valerio; Hartshorn, Michael J.; Mooij, Wijnand T. M.; Murray, Christopher W.; Taylor, Richard D.; Watson, Paul
Journal of Chemical Information and Computer Sciences (2004), 44 (3), 793-806CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
This study addresses a no. of topical issues around the use of protein-ligand docking in virtual screening. We show that, for the validation of such methods, it is key to use focused libraries (contg. compds. with one-dimensional properties, similar to the actives), rather than "random" or "drug-like" libraries to test the actives against. We also show that, to obtain good enrichments, the docking program needs to produce reliable binding modes. We demonstrate how pharmacophores can be used to guide the dockings and improve enrichments, and we compare the performance of three consensus-ranking protocols against ranking based on individual scoring functions. Finally, we show that protein-ligand docking can be an effective aid in the screening for weak, fragment-like binders, which has rapidly become a popular strategy for hit identification. All results presented are based on carefully constructed virtual screening expts. against four targets, using the protein-ligand docking program GOLD.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXhtVGhs70%253D&md5=b0cd9e9a04ddebe401dee9744e2184cb
18
Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A. The maximal affinity of ligands Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 9997– 10002
Google Scholar
18
The maximal affinity of ligands
Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A.
Proceedings of the National Academy of Sciences of the United States of America (1999), 96 (18), 9997-10002CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
We explore the question of what are the best ligands for macromol. targets. A survey of exptl. data on a large no. of the strongest-binding ligands indicates that the free energy of binding increases with the no. of nonhydrogen atoms with an initial slope of ≈-1.5 kcal/mol (1 cal = 4.18 J) per atom. For ligands that contain more than 15 nonhydrogen atoms, the free energy of binding increases very little with relative mol. mass. This nonlinearity is largely ascribed to nonthermodynamic factors. An anal. of the dominant interactions suggests that van der Waals interactions and hydrophobic effects provide a reasonable basis for understanding binding affinities across the entire set of ligands. Interesting outliers that bind unusually strongly on a per atom basis include metal ions, covalently attached ligands, and a few well known complexes such as biotin-avidin.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK1MXlvFehu7s%253D&md5=46fe894b016f3041831a2f0b71f812b1
19
Fan, H.; Irwin, J. J.; Webb, B. M.; Klebe, G.; Shoichet, B. K.; Sali, A. Molecular Docking Screens Using Comparative Models of Proteins J. Chem. Inf. Model. 2009, 49, 2512– 2527
Google Scholar
19
Molecular Docking Screens Using Comparative Models of Proteins
Fan, Hao; Irwin, John J.; Webb, Benjamin M.; Klebe, Gerhard; Shoichet, Brian K.; Sali, Andrej
Journal of Chemical Information and Modeling (2009), 49 (11), 2512-2527CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Two orders of magnitude more protein sequences can be modeled by comparative modeling than have been detd. by X-ray crystallog. and NMR spectroscopy. Investigators have nevertheless been cautious about using comparative models for ligand discovery because of concerns about model errors. We suggest how to exploit comparative models for mol. screens, based on docking against a wide range of crystallog. structures and comparative models with known ligands. To account for the variation in the ligand-binding pocket as it binds different ligands, we calc. "consensus" enrichment by ranking each library compd. by its best docking score against all available comparative models and/or modeling templates. For the majority of the targets, the consensus enrichment for multiple models was better than or comparable to that of the holo and apo X-ray structures. Even for single models, the models are significantly more enriching than the template structure if the template is paralogous and shares more than 25% sequence identity with the target.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXht1yntL7M&md5=11bd5284c6b43592906e45b6aea3019d
20
Repasky, M. P.; Murphy, R. B.; Banks, J. L.; Greenwood, J. R.; Tubert-Brohman, I.; Bhat, S.; Friesner, R. A. Docking performance of the glide program as evaluated on the Astex and DUD datasets: a complete set of glide SP results and selected results for a new scoring function integrating WaterMap and glide J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9575-9
Google Scholar
There is no corresponding record for this reference.
21
Brozell, S. R.; Mukherjee, S.; Balius, T. E.; Roe, D. R.; Case, D. A.; Rizzo, R. C. Evaluation of DOCK 6 as a pose generation and database enrichment tool J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9565-y
Google Scholar
There is no corresponding record for this reference.
22
Neves, M. A.; Totrov, M.; Abagyan, R. Docking and scoring with ICM: the benchmarking results and strategies for improvement J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9547-0
Google Scholar
There is no corresponding record for this reference.
23
Spitzer, R.; Jain, A. N. Surflex-Dock: docking benchmarks and real-world application J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-011-9533-y
Google Scholar
There is no corresponding record for this reference.
24
Schneider, N.; Hindle, S.; Lange, G.; Klein, R.; Albrecht, J.; Briem, H.; Beyer, K.; Claussen, H.; Gastreich, M.; Lemmen, C.; Rarey, M. Substantial improvements in large-scale redocking and screening using the novel HYDE scoring function J. Comput.-Aided Mol. Des. 2011, DOI: 10.1007/s10822-011-9531-0
Google Scholar
There is no corresponding record for this reference.
25
Liebeschuetz, J. W.; Cole, J. C.; Korb, O. Pose prediction and virtual screening performance of GOLD scoring functions in a standardized test J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9551-4
Google Scholar
There is no corresponding record for this reference.
26
Novikov, F. N.; Stroylov, V. S.; Zeifman, A. A.; Stroganov, O. V.; Kulkov, V.; Chilov, G. G. Lead Finder docking and virtual screening evaluation with Astex and DUD test sets J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9549-y
Google Scholar
There is no corresponding record for this reference.
27
Good, A. C.; Oprea, T. I. Optimization of CAMD techniques 3. Virtual screening enrichment studies: a help or hindrance in tool selection? J. Comput.-Aided Mol. Des. 2008, 22, 169– 178
Google Scholar
27
Optimization of CAMD techniques 3. Virtual screening enrichment studies: a help or hindrance in tool selection?
Good, Andrew C.; Oprea, Tudor I.
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 169-178CODEN: JCADEQ; ISSN:0920-654X. (Springer)
Over the last few years many articles have been published in an attempt to provide performance benchmarks for virtual screening tools. While this research has imparted useful insights, the myriad variables controlling said studies place significant limits on results interpretability. Here we investigate the effects of these variables, including anal. of calcn. setup variation, the effect of target choice, active/decoy set selection (with particular emphasis on the effect of analog bias) and enrichment data interpretation. In addn. the optimization of the publicly available DUD benchmark sets through analog bias removal is discussed, as is their augmentation through the addn. of large diverse data sets collated using WOMBAT.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsLo%253D&md5=459cd3d09a88d5afc04a4a49e172fb76
28
Mackey, M. D.; Melville, J. L. Better than random? The chemotype enrichment problem J. Chem. Inf. Model. 2009, 49, 1154– 1162
Google Scholar
28
Better than Random? The Chemotype Enrichment Problem
Mackey, Mark D.; Melville, James L.
Journal of Chemical Information and Modeling (2009), 49 (5), 1154-1162CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Chemotype enrichment is increasingly recognized as an important measure of virtual screening performance. However, little attention has been paid to producing metrics which can quantify chemotype retrieval. Here, we examine two different protocols for analyzing chemotype retrieval: "cluster averaging", where the contribution of each active to the scoring metric is proportional to the no. of other actives with the same chemotype, and "first found", where only the first active for a given chemotype contributes to the score. We demonstrate that this latter anal., common in the qual. anal. used in the current literature, has important drawbacks when combined with quant. metrics.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXltVCgs78%253D&md5=51b6efb624c3a383b7546997c11ab1e0
29
Hawkins, P. C.; Warren, G. L.; Skillman, A. G.; Nicholls, A. How to do an evaluation: pitfalls and traps J. Comput.-Aided Mol. Des. 2008, 22, 179– 190
Google Scholar
29
How to do an evaluation: pitfalls and traps
Hawkins, Paul C. D.; Warren, Gregory L.; Skillman, A. Geoffrey; Nicholls, Anthony
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 179-190CODEN: JCADEQ; ISSN:0920-654X. (Springer)
The recent literature is replete with papers evaluating computational tools (often those operating on 3D structures) for their performance in a certain set of tasks. Most commonly these papers compare a no. of docking tools for their performance in cognate re-docking (pose prediction) and/or virtual screening. Related papers have been published on ligand-based tools: pose prediction by conformer generators and virtual screening using a variety of ligand-based approaches. The reliability of these comparisons is critically affected by a no. of factors usually ignored by the authors, including bias in the datasets used in virtual screening, the metrics used to assess performance in virtual screening and pose prediction and errors in crystal structures used.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsLk%253D&md5=3e176941c4a526f3a97519ce02007792
30
Irwin, J. J. Community benchmarks for virtual screening J. Comput.-Aided Mol. Des. 2008, 22, 193– 199
Google Scholar
30
Community benchmarks for virtual screening
Irwin, John J.
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 193-199CODEN: JCADEQ; ISSN:0920-654X. (Springer)
Ligand enrichment among top-ranking hits is a key metric of virtual screening. To avoid bias, decoys should resemble ligands phys., so that enrichment is not attributable to simple differences of gross features. We therefore created a directory of useful decoys (DUD) by selecting decoys that resembled annotated ligands phys. but not topol. to benchmark docking performance. DUD has 2950 annotated ligands and 95,316 property-matched decoys for 40 targets. It is by far the largest and most comprehensive public data set for benchmarking virtual screening programs that I am aware of. This paper outlines several ways that DUD can be improved to provide better telemetry to investigators seeking to understand both the strengths and the weaknesses of current docking methods. I also highlight several pitfalls for the unwary: a risk of over-optimization, questions about chem. space, and the proper scope for using DUD. Careful attention to both the compn. of benchmarks and how they are used is essential to avoid being misled by overfitting and bias.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsb8%253D&md5=4972b768c915bc3e7ee05210dcf1b7a8
31
Mysinger, M. M.; Shoichet, B. K. Rapid context-dependent ligand desolvation in molecular docking J. Chem. Inf. Model. 2010, 50, 1561– 1573
Google Scholar
31
Rapid Context-Dependent Ligand Desolvation in Molecular Docking
Mysinger, Michael M.; Shoichet, Brian K.
Journal of Chemical Information and Modeling (2010), 50 (9), 1561-1573CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
In structure-based screens for new ligands, a mol. docking algorithm must rapidly score many mols. in multiple configurations, accounting for both the ligand's interactions with receptor and its competing interactions with solvent. Here the authors explore a context-dependent ligand desolvation scoring term for mol. docking. The authors relate the Generalized-Born effective Born radii for every ligand atom to a fractional desolvation and then use this fraction to scale an atom-by-atom decompn. of the full transfer free energy. The fractional desolvation is precomputed on a scoring grid by numerically integrating over the vol. of receptor proximal to a ligand atom, weighted by distance. To test this method's performance, the authors dock ligands vs. property-matched decoys over 40 DUD targets. Context-dependent desolvation better enriches ligands compared to both the raw full transfer free energy penalty and compared to ignoring desolvation altogether, though the improvement is modest. More compellingly, the new method improves docking performance across receptor types. Thus, whereas entirely ignoring desolvation works best for charged sites and overpenalizing with full desolvation works well for neutral sites, the phys. more correct context-dependent ligand desolvation is competitive across both types of targets. The method also reliably discriminates ligands from highly charged mols., where ignoring desolvation performs poorly. Since this context-dependent ligand desolvation may be precalcd., it improves docking reliability with minimal cost to calcn. time and may be readily incorporated into any physics-based docking program.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXhtVGrsLvI&md5=5124d2c4618e07b4f1e41609c9e61362
32
Vogel, S. M.; Bauer, M. R.; Boeckler, F. M. DEKOIS: demanding evaluation kits for objective in silico screening—a versatile tool for benchmarking docking programs and scoring functions J. Chem. Inf. Model. 2011, 51, 2650– 2665
Google Scholar
32
DEKOIS: Demanding Evaluation Kits for Objective in Silico Screening - A Versatile Tool for Benchmarking Docking Programs and Scoring Functions
Vogel, Simon M.; Bauer, Matthias R.; Boeckler, Frank M.
Journal of Chemical Information and Modeling (2011), 51 (10), 2650-2665CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
For widely applied in silico screening techniques success depends on the rational selection of an appropriate method. We herein present a fast, versatile, and robust method to construct demanding evaluation kits for objective in silico screening (DEKOIS). This automated process enables creating tailor-made decoy sets for any given sets of bioactives. It facilitates a target-dependent validation of docking algorithms and scoring functions helping to save time and resources. We have developed metrics for assessing and improving decoy set quality and employ them to investigate how decoy embedding affects docking. We demonstrate that screening performance is target-dependent and can be impaired by latent actives in the decoy set (LADS) or enhanced by poor decoy embedding. The presented method allows extending and complementing the collection of publicly available high quality decoy sets toward new target space. All present and future DEKOIS data sets will be made accessible at www.dekois.com.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtValsbnN&md5=e3f59bcd3f898dce1c116a325e914861
33
Wallach, I.; Lilien, R. Virtual decoy sets for molecular docking benchmarks J. Chem. Inf. Model. 2011, 51, 196– 202
Google Scholar
33
Virtual Decoy Sets for Molecular Docking Benchmarks
Wallach, Izhar; Lilien, Ryan
Journal of Chemical Information and Modeling (2011), 51 (2), 196-202CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Virtual docking algorithms are often evaluated on their ability to sep. active ligands from decoy mols. The current state-of-the-art benchmark, the Directory of Useful Decoys (DUD), minimizes bias by including decoys from a library of synthetically feasible mols. that are phys. similar yet chem. dissimilar to the active ligands. We show that by ignoring synthetic feasibility, we can compile a benchmark that is comparable to the DUD and less biased with respect to phys. similarity.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXoslel&md5=3b58c7641cb2c665ce381081d8b1bf41
34
Gatica, E. A.; Cavasotto, C. N. Ligand and decoy sets for docking to G protein-coupled receptors J. Chem. Inf. Model. 2012, 52, 1– 6
Google Scholar
34
Ligand and Decoy Sets for Docking to G Protein-Coupled Receptors
Gatica, Edgar A.; Cavasotto, Claudio N.
Journal of Chemical Information and Modeling (2012), 52 (1), 1-6CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
We compiled a G protein-coupled receptor (GPCR) ligand library (GLL) for 147 targets, selecting for each ligand 39 decoy mols., collected in the GPCR Decoy Database (GDD). Decoys were chosen ensuring a ligand-decoy similarity of six phys. properties, while enforcing ligand-decoy chem. dissimilarity. The performance in docking of the GDD was evaluated on 19 GPCRs, showing a marked decrease in enrichment compared to bias-uncorrected decoy sets. Both the GLL and GDD are freely available for the scientific community.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1ajsb3P&md5=a51e1b60afda728614a3bc1ceed79cfd
35
Cereto-Massague, A.; Guasch, L.; Valls, C.; Mulero, M.; Pujadas, G.; Garcia-Vallve, S. DecoyFinder: an easy-to-use python GUI application for building target-specific decoy sets Bioinformatics 2012, 28, 1661– 1662
Google Scholar
There is no corresponding record for this reference.
36
Rohrer, S. G.; Baumann, K. Maximum unbiased validation (MUV) data sets for virtual screening based on PubChem bioactivity data J. Chem. Inf. Model. 2009, 49, 169– 184
Google Scholar
36
Maximum Unbiased Validation (MUV) Data Sets for Virtual Screening Based on PubChem Bioactivity Data
Rohrer, Sebastian G.; Baumann, Knut
Journal of Chemical Information and Modeling (2009), 49 (2), 169-184CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Refined nearest neighbor anal. was recently introduced for the anal. of virtual screening benchmark data sets. It constitutes a technique from the field of spatial statistics and provides a math. framework for the nonparametric anal. of mapped point patterns. Here, refined nearest neighbor anal. is used to design benchmark data sets for virtual screening based on PubChem bioactivity data. A workflow is devised that purges data sets of compds. active against pharmaceutically relevant targets from unselective hits. Topol. optimization using exptl. design strategies monitored by refined nearest neighbor anal. functions is applied to generate corresponding data sets of actives and decoys that are unbiased with regard to analog bias and artificial enrichment. These data sets provide a tool for Maximum Unbiased Validation (MUV) of virtual screening methods. The data sets and a software package implementing the MUV design workflow are freely available.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXptlOhtQ%253D%253D&md5=17805f49d995c4c082c3d07be7145a2d
37
Ripphausen, P.; Wassermann, A. M.; Bajorath, J. REPROVIS-DB: a benchmark system for ligand-based virtual screening derived from reproducible prospective applications J. Chem. Inf. Model. 2011, 51, 2467– 2473
Google Scholar
37
REPROVIS-DB: A Benchmark System for Ligand-Based Virtual Screening Derived from Reproducible Prospective Applications
Ripphausen, Peter; Wassermann, Anne Mai; Bajorath, Jurgen
Journal of Chemical Information and Modeling (2011), 51 (10), 2467-2473CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Benchmark calcns. are essential for the evaluation of virtual screening (VS) methods. Typically, classes of known active compds. taken from the medicinal chem. literature are divided into ref. mols. (search templates) and potential hits that are added to background databases assumed to consist of compds. not sharing this activity. Then VS calcns. are carried out, and the recall of known active compds. is detd. However, conventional benchmarking is affected by a no. of problems that reduce its value for method evaluation. In addn. to often insufficient statistical validation and the lack of generally accepted evaluation stds., the artificial nature of typical benchmark settings is often criticized. Retrospective benchmark calcns. generally overestimate the potential of VS methods and do not scale with their performance in prospective applications. In order to provide addnl. opportunities for benchmarking that more closely resemble practical VS conditions, we have designed a publicly available compd. database (DB) of reproducible virtual screens (REPROVIS-DB) that organizes information from successful ligand-based VS applications including ref. compds., screening databases, compd. selection criteria, and exptl. confirmed hits. Using the currently available 25 hand-selected compd. data sets, one can attempt to reproduce successful virtual screens with other than the originally applied methods and assess their potential for practical applications.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXht1ait7fJ&md5=8439a3f6471f43afe159d08c28a5bc14
38
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– 1107
Google Scholar
38
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
39
Bemis, G. W.; Murcko, M. A. The properties of known drugs. 1. Molecular frameworks J. Med. Chem. 1996, 39, 2887– 2893
Google Scholar
39
The Properties of Known Drugs. 1. Molecular Frameworks
Bemis, Guy W.; Murcko, Mark A.
Journal of Medicinal Chemistry (1996), 39 (15), 2887-2893CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
To better understand the common features present in drug mols., we use shape description methods to analyze a database of com. available drugs and prep. a list of common drug shapes. A useful way of organizing this structural data is to group the atoms of each drug mol. into ring, linker, framework, and side chain atoms. On the basis of the two-dimensional mol. structures (without regard to atom type, hybridization, and bond order), there are 1179 different frameworks among the 5120 compds. analyzed. However, the shapes of half of the drugs in the database are described by the 32 most frequently occurring frameworks. This suggests that the diversity of shapes in the set of known drugs is extremely low. In our second method of anal., in which atom type, hybridization, and bond order are considered, more diversity is seen; there are 2506 different frameworks among the 5120 compds. in the database, and the most frequently occurring 42 frameworks account for only one-fourth of the drugs. We discuss the possible interpretations of these findings and the way they may be used to guide future drug discovery research.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK28XjvVejtro%253D&md5=5e2c4fdfea9434456a0cca83de4185b3
40
Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank Nucleic Acids Res. 2000, 28, 235– 242
Google Scholar
40
The Protein Data Bank
Berman, Helen M.; Westbrook, John; Feng, Zukang; Gilliland, Gary; Bhat, T. N.; Weissig, Helge; Shindyalov, Ilya N.; Bourne, Philip E.
Nucleic Acids Research (2000), 28 (1), 235-242CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
The Protein Data Bank (PDB; http://www.rcsb.org/pdb/)is the single worldwide archive of structural data of biol. macromols. This paper describes the goals of the PDB, the systems in place for data deposition and access, how to obtain further information, and near-term plans for the future development of the resource.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXhvVKjt7w%253D&md5=227fb393f754be2be375ab727bfd05dc
41
Apweiler, R.; Bairoch, A.; Wu, C. H.; Barker, W. C.; Boeckmann, B.; Ferro, S.; Gasteiger, E.; Huang, H.; Lopez, R.; Magrane, M.; Martin, M. J.; Natale, D. A.; O’Donovan, C.; Redaschi, N.; Yeh, L. S. UniProt: The Universal Protein Knowledgebase Nucleic Acids Res. 2004, 32, D115– D119
Google Scholar
41
UniProt: the universal protein knowledgebase
Apweiler, Rolf; Bairoch, Amos; Wu, Cathy H.; Barker, Winona C.; Boeckmann, Brigitte; Ferro, Serenella; Gasteiger, Elisabeth; Huang, Hongzhan; Lopez, Rodrigo; Magrane, Michele; Martin, Maria J.; Natale, Darren A.; O'Donovan, Claire; Redaschi, Nicole; Yeh, Lai-Su L.
Nucleic Acids Research (2004), 32 (Database), D115-D119CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
To provide the scientific community with a single, centralized, authoritative resource for protein sequences and functional information, the Swiss-Prot, TrEMBL and PIR protein database activities have united to form the Universal Protein Knowledgebase (UniProt) consortium. Our mission is to provide a comprehensive, fully classified, richly and accurately annotated protein sequence knowledgebase, with extensive cross-refs. and query interfaces. The central database will have two sections, corresponding to the familiar Swiss-Prot (fully manually curated entries) and TrEMBL (enriched with automated classification, annotation and extensive cross-refs.). For convenient sequence searches, UniProt also provides several non-redundant sequence databases. The UniProt NREF (UniRef) databases provide representative subsets of the knowledgebase suitable for efficient searching. The comprehensive UniProt Archive (UniParc) is updated daily from many public source databases. The UniProt databases can be accessed online (http://www.uniprot.org) or downloaded in several formats (ftp://ftp.uniprot.org/pub). The scientific community is encouraged to submit data for inclusion in UniProt.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3sXhtVSru7vK&md5=4b692ffbb5c27bd9b4d2a8d0809e84e1
42
Irwin, J. J.; Shoichet, B. K.; Mysinger, M. M.; Huang, N.; Colizzi, F.; Wassam, P.; Cao, Y. Automated docking screens: a feasibility study J. Med. Chem. 2009, 52, 5712– 5720
Google Scholar
42
Automated Docking Screens: A Feasibility Study
Irwin, John J.; Shoichet, Brian K.; Mysinger, Michael M.; Huang, Niu; Colizzi, Francesco; Wassam, Pascal; Cao, Yiqun
Journal of Medicinal Chemistry (2009), 52 (18), 5712-5720CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Mol. docking is the most practical approach to leverage protein structure for ligand discovery, but the technique retains important liabilities that make it challenging to deploy on a large scale. We have therefore created an expert system, DOCK Blaster, to investigate the feasibility of full automation. The method requires a PDB code, sometimes with a ligand structure, and from that alone can launch a full screen of large libraries. A crit. feature is self-assessment, which ests. the anticipated reliability of the automated screening results using pose fidelity and enrichment. Against common benchmarks, DOCK Blaster recapitulates the crystal ligand pose within 2 Å rmsd 50-60% of the time; inferior to an expert, but respectable. Half the time the ligand also ranked among the top 5% of 100 phys. matched decoys chosen on the fly. Further tests were undertaken culminating in a study of 7755 eligible PDB structures. In 1398 cases, the redocked ligand ranked in the top 5% of 100 property-matched decoys while also posing within 2 Å rmsd, suggesting that unsupervised prospective docking is viable. DOCK Blaster is available at http://blaster.docking.org.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXhtVOjurjN&md5=f302db1f77b87f26d1fa2fece55bca96
43
Powers, R. A.; Morandi, F.; Shoichet, B. K. Structure-based discovery of a novel, noncovalent inhibitor of AmpC beta-lactamase Structure 2002, 10, 1013– 1023
Google Scholar
43
Structure-Based Discovery of a Novel, Noncovalent Inhibitor of AmpC β-Lactamase
Powers, Rachel A.; Morandi, Federica; Shoichet, Brian K.
Structure (Cambridge, MA, United States) (2002), 10 (7), 1013-1023CODEN: STRUE6; ISSN:0969-2126. (Cell Press)
β-Lactamases are the most widespread resistance mechanisms to β-lactam antibiotics, and there is a pressing need for novel, non-β-lactam drugs. A database of over 200,000 compds. was docked to the active site of AmpC β-lactamase to identify potential inhibitors. Fifty-six compds. were tested, and three had Ki values of 650 μM or better. The best of these, 3-[(4-chloroanilino)sulfonyl]thiophene-2-carboxylic acid, was a competitive noncovalent inhibitor (Ki = 26 μM), which also reversed resistance to β-lactams in bacteria expressing AmpC. The structure of AmpC in complex with this compd. was detd. by x-ray crystallog. to 1.94 A and reveals that the inhibitor interacts with key active-site residues in sites targeted in the docking calcn. Indeed, the exptl. detd. conformation of the inhibitor closely resembles the prediction. The structure of the enzyme-inhibitor complex presents an opportunity to improve binding affinity in a novel series of inhibitors discovered by structure-based methods.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38XlsVyjsrk%253D&md5=5f539484d08191f0c65626493fc1b262
44
Carlsson, J.; Yoo, L.; Gao, Z. G.; Irwin, J. J.; Shoichet, B. K.; Jacobson, K. A. Structure-based discovery of A2A adenosine receptor ligands J. Med. Chem. 2010, 53, 3748– 3755
Google Scholar
44
Structure-Based Discovery of A2A Adenosine Receptor Ligands
Carlsson, Jens; Yoo, Lena; Gao, Zhan-Guo; Irwin, John J.; Shoichet, Brian K.; Jacobson, Kenneth A.
Journal of Medicinal Chemistry (2010), 53 (9), 3748-3755CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
The recent detn. of X-ray structures of pharmacol. relevant GPCRs has made these targets accessible to structure-based ligand discovery. Here we explore whether novel chemotypes may be discovered for the A2A adenosine receptor, based on complementarity to its recently detd. structure. The A2A adenosine receptor signals in the periphery and the CNS, with agonists explored as anti-inflammatory drugs and antagonists explored for neurodegenerative diseases. We used mol. docking to screen a 1.4 million compd. database against the X-ray structure computationally and tested 20 high-ranking, previously unknown mols. exptl. Of these 35% showed substantial activity with affinities between 200 nM and 9 μM. For the most potent of these new inhibitors, over 50-fold specificity was obsd. for the A2A vs. the related A1 and A3 subtypes. These high hit rates and affinities at least partly reflect the bias of com. libraries toward GPCR-like chemotypes, an issue that we attempt to investigate quant. Despite this bias, many of the most potent new ligands were novel, dissimilar from known ligands, providing new lead structures for modulation of this medically important target.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXkvFaqsL8%253D&md5=c36c941d52d2cec06387d79c4c423d46
45
Carlsson, J.; Coleman, R. G.; Setola, V.; Irwin, J. J.; Fan, H.; Schlessinger, A.; Sali, A.; Roth, B. L.; Shoichet, B. K. Ligand discovery from a dopamine D3 receptor homology model and crystal structure Natre Chem. Biol. 2011, 7, 769– 778
Google Scholar
45
Ligand discovery from a dopamine D3 receptor homology model and crystal structure
Carlsson, Jens; Coleman, Ryan G.; Setola, Vincent; Irwin, John J.; Fan, Hao; Schlessinger, Avner; Sali, Andrej; Roth, Bryan L.; Shoichet, Brian K.
Nature Chemical Biology (2011), 7 (11), 769-778CODEN: NCBABT; ISSN:1552-4450. (Nature Publishing Group)
G protein-coupled receptors (GPCRs) are intensely studied as drug targets and for their role in signaling. With the detn. of the first crystal structures, interest in structure-based ligand discovery increased. Unfortunately, for most GPCRs no exptl. structures are available. The detn. of the D3 receptor structure and the challenge to the community to predict it enabled a fully prospective comparison of ligand discovery from a modeled structure vs. that of the subsequently released crystal structure. Over 3.3 million mols. were docked against a homol. model, and 26 of the highest ranking were tested for binding. Six had affinities ranging from 0.2 to 3.1 μM. Subsequently, the crystal structure was released and the docking screen repeated. Of the 25 compds. selected, five had affinities ranging from 0.3 to 3.0 μM. One of the new ligands from the homol. model screen was optimized for affinity to 81 nM. The feasibility of docking screens against modeled GPCRs more generally is considered.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtFylsLvP&md5=7242aa512079a348678b309d352ecdb9
46
Irwin, J. J.; Shoichet, B. K. ZINC—a free database of commercially available compounds for virtual screening J. Chem. Inf. Model. 2005, 45, 177– 182
Google Scholar
46
ZINC - A Free Database of Commercially Available Compounds for Virtual Screening
Irwin, John J.; Shoichet, Brian K.
Journal of Chemical Information and Computer Sciences (2005), 45 (1), 177-182CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
A crit. barrier to entry into structure-based virtual screening is the lack of a suitable, easy to access database of purchasable compds. We have therefore prepd. a library of 727 842 mols., each with 3D structure, using catalogs of compds. from vendors (the size of this library continues to grow). The mols. have been assigned biol. relevant protonation states and are annotated with properties such as mol. wt., calcd. LogP, and no. of rotatable bonds. Each mol. in the library contains vendor and purchasing information and is ready for docking using a no. of popular docking programs. Within certain limits, the mols. are prepd. in multiple protonation states and multiple tautomeric forms. In one format, multiple conformations are available for the mols. This database is available for free download (http://zinc.docking.org) in several common file formats including SMILES, mol2, 3D SDF, and DOCK flexibase format. A Web-based query tool incorporating a mol. drawing interface enables the database to be searched and browsed and subsets to be created. Users can process their own mols. by uploading them to a server. Our hope is that this database will bring virtual screening libraries to a wide community of structural biologists and medicinal chemists.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXhtVOjt77J&md5=e3892b7dc8608b17a3e63541a5ed60e6
47
Velankar, S.; McNeil, P.; Mittard-Runte, V.; Suarez, A.; Barrell, D.; Apweiler, R.; Henrick, K. E-MSD: an integrated data resource for bioinformatics Nucleic Acids Res. 2005, 33, D262– 265
Google Scholar
There is no corresponding record for this reference.
48
Hawkins, P. C.; Skillman, A. G.; Nicholls, A. Comparison of shape-matching and docking as virtual screening tools J. Med. Chem. 2007, 50, 74– 82
Google Scholar
48
Comparison of Shape-Matching and Docking as Virtual Screening Tools
Hawkins, Paul C. D.; Skillman, A. Geoffrey; Nicholls, Anthony
Journal of Medicinal Chemistry (2007), 50 (1), 74-82CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Ligand docking is a widely used approach in virtual screening. In recent years a large no. of publications have appeared in which docking tools are compared and evaluated for their effectiveness in virtual screening against a wide variety of protein targets. These studies have shown that the effectiveness of docking in virtual screening is highly variable due to a large no. of possible confounding factors. Another class of method that has shown promise in virtual screening is the shape-based, ligand-centric approach. Several direct comparisons of docking with the shape-based tool ROCS have been conducted using data sets from some of these recent docking publications. The results show that a shape-based, ligand-centric approach is more consistent than, and often superior to, the protein-centric approach taken by docking.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28Xhtlansb%252FF&md5=6f97f5c0cc092b4e225f7c2656c1bcf6
49
Teotico, D. G.; Babaoglu, K.; Rocklin, G. J.; Ferreira, R. S.; Giannetti, A. M.; Shoichet, B. K. Docking for fragment inhibitors of AmpC beta-lactamase Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 7455– 7460
Google Scholar
49
Docking for fragment inhibitors of AmpC 83-lactamase
Teotico, Denise G.; Babaoglu, Kerim; Rocklin, Gabriel J.; Ferreira, Rafaela S.; Giannetti, Anthony M.; Shoichet, Brian K.
Proceedings of the National Academy of Sciences of the United States of America (2009), 106 (18), 7455-7460CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
Fragment screens for new ligands have had wide success, not withstanding their constraint to libraries of 1000-10,000 mols. Larger libraries would be addressable were mol. docking reliable for fragment screens, but this has not been widely accepted. To investigate docking's ability to prioritize fragments, a library of >137,000 such mols. were docked against the structure of β-lactamase. Forty-eight fragments highly ranked by docking were acquired and tested; 23 had Ki values ranging from 0.7 to 9.2 mM. X-ray crystal structures of the enzyme-bound complexes were detd. for 8 of the fragments. For 4, the correspondence between the predicted and exptl. structures was high (RMSD between 1.2 and 1.4 Å), whereas for another 2, the fidelity was lower but retained most key interactions (RMSD 2.4-2.6 Å). Two of the 8 fragments adopted very different poses in the active site owing to enzyme conformational changes. The 48% hit rate of the fragment docking compares very favorably with "lead-like" docking and high-throughput screening against the same enzyme. To understand this, we investigated the occurrence of the fragment scaffolds among larger, lead-like mols. Approx. 1% of com. available fragments contain these inhibitors whereas only 10-7% of lead-like mols. do. This suggests that many more chemotypes and combinations of chemotypes are present among fragments than are available among lead-like mols., contributing to the higher hit rates. The ability of docking to prioritize these fragments suggests that the technique can be used to exploit the better chemotype coverage that exists at the fragment level.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXmt1Khtrg%253D&md5=d2df01432d5381c226e8aa7e008457b5
50
Tondi, D.; Morandi, F.; Bonnet, R.; Costi, M. P.; Shoichet, B. K. Structure-based optimization of a non-beta-lactam lead results in inhibitors that do not up-regulate beta-lactamase expression in cell culture J. Am. Chem. Soc. 2005, 127, 4632– 4639
Google Scholar
50
Structure-Based Optimization of a Non-β-lactam Lead Results in Inhibitors That Do Not Up-Regulate β-Lactamase Expression in Cell Culture
Tondi, Donatella; Morandi, Federica; Bonnet, Richard; Costi, M. Paola; Shoichet, Brian K.
Journal of the American Chemical Society (2005), 127 (13), 4632-4639CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
Bacterial expression of β-lactamases is the most widespread resistance mechanism to β-lactam antibiotics, such as penicillins and cephalosporins. There is a pressing need for novel, non-β-lactam inhibitors of these enzymes. One previously discovered novel inhibitor of the β-lactamase AmpC (I) has several favorable properties: it is chem. dissimilar to β-lactams and is a noncovalent, competitive inhibitor of the enzyme. However, at 26 μM its activity is modest. Using the x-ray structure of the AmpC/I complex as a template, 14 analogs were designed and synthesized. The most active of these, (II), had a Ki of 1 μM, 26-fold better than the lead. To understand the origins of this improved activity, the structures of AmpC in complex with compd. II and an analog were detd. by x-ray crystallog. to 1.97 and 1.96 Å, resp. II was active in cell culture, reversing resistance to the third generation cephalosporin ceftazidime in bacterial pathogens expressing AmpC. In contrast to β-lactam-based inhibitors clavulanate and cefoxitin, compd. II did not up-regulate β-lactamase expression in cell culture but simply inhibited the enzyme expressed by the resistant bacteria. Its escape from this resistance mechanism derives from its dissimilarity to β-lactam antibiotics.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXit1Krsbk%253D&md5=ddad35ac3e6fb96cf65e12cd807f271c
51
Graves, A. P.; Brenk, R.; Shoichet, B. K. Decoys for docking J. Med. Chem. 2005, 48, 3714– 3728
Google Scholar
51
Decoys for Docking
Graves, Alan P.; Brenk, Ruth; Shoichet, Brian K.
Journal of Medicinal Chemistry (2005), 48 (11), 3714-3728CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
Mol. docking is widely used to predict novel lead compds. for drug discovery. Success depends on the quality of the docking scoring function, among other factors. An imperfect scoring function can mislead by predicting incorrect ligand geometries or by selecting nonbinding mols. over true ligands. These false-pos. hits may be considered "decoys". Although these decoys are frustrating, they potentially provide important tests for a docking algorithm; the more subtle the decoy, the more rigorous the test. Indeed, decoy databases have been used to improve protein structure prediction algorithms and protein-protein docking algorithms. Here, we describe 20 geometric decoys in five enzymes and 166 "hit list" decoys-i.e., mols. predicted to bind by our docking program that were tested and found not to do so - for β-lactamase and two cavity sites in lysozyme. Esp. in the cavity sites, which are very simple, these decoys highlight particular weaknesses in our scoring function. We also consider the performance of five other widely used docking scoring functions against our geometric and hit list decoys. Intriguingly, whereas many of these other scoring functions performed better on the geometric decoys, they typically performed worse on the hit list decoys, often highly ranking mols. that seemed to poorly complement the model sites. Several of these "hits" from the other scoring functions were tested exptl. and found, in fact, to be decoys. Collectively, these decoys provide a tool for the development and improvement of mol. docking scoring functions. Such improvements may, in turn, be rapidly tested exptl. against these and related exptl. systems, which are well-behaved in assays and for structure detn.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXjvFKjsbs%253D&md5=1797b501c1777dc94553701d45c822f0
52
Hawkins, P. C.; Skillman, A. G.; Warren, G. L.; Ellingson, B. A.; Stahl, M. T. Conformer generation with OMEGA: algorithm and validation using high quality structures from the Protein Data Bank and Cambridge Structural Database J. Chem. Inf. Model. 2010, 50, 572– 584
Google Scholar
52
Conformer Generation with OMEGA: Algorithm and Validation Using High Quality Structures from the Protein Databank and Cambridge Structural Database
Hawkins, Paul C. D.; Skillman, A. Geoffrey; Warren, Gregory L.; Ellingson, Benjamin A.; Stahl, Matthew T.
Journal of Chemical Information and Modeling (2010), 50 (4), 572-584CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
Here, we present the algorithm and validation for OMEGA, a systematic, knowledge-based conformer generator. The algorithm consists of three phases: assembly of an initial 3D structure from a library of fragments; exhaustive enumeration of all rotatable torsions using values drawn from a knowledge-based list of angles, thereby generating a large set of conformations; and sampling of this set by geometric and energy criteria. Validation of conformer generators like OMEGA has often been undertaken by comparing computed conformer sets to exptl. mol. conformations from crystallog., usually from the Protein Databank (PDB). Such an approach is fraught with difficulty due to the systematic problems with small mol. structures in the PDB. Methods are presented to identify a diverse set of small mol. structures from cocomplexes in the PDB that has maximal reliability. A challenging set of 197 high quality, carefully selected ligand structures from well-solved models was obtained using these methods. This set will provide a sound basis for comparison and validation of conformer generators in the future. Validation results from this set are compared to the results using structures of a set of druglike mols. extd. from the Cambridge Structural Database (CSD). OMEGA is found to perform very well in reproducing the crystallog. conformations from both these data sets using two complementary metrics of success.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXjtlaisrY%253D&md5=fb87ecc9c51eddef63b41fffcd9babee
53
Jain, A. N. Bias, reporting, and sharing: computational evaluations of docking methods J. Comput.-Aided Mol. Des. 2008, 22, 201– 212
Google Scholar
53
Bias, reporting, and sharing: computational evaluations of docking methods
Jain, Ajay N.
Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 201-212CODEN: JCADEQ; ISSN:0920-654X. (Springer)
Computational methods for docking ligands to protein binding sites have become ubiquitous in drug discovery. Despite the age of the field, no stds. have been established with respect to methodol. evaluation of docking accuracy, virtual screening utility, or scoring accuracy. There are crit. issues relating to data sharing, data set design and prepn., and statistical reporting that have an impact on the degree to which a report will translate into real-world performance. These issues also have an impact on whether there is a transparent relationship between methodol. changes and reported performance improvements. This paper presents detailed examples of pitfalls in each area and makes recommendations as to best practices.
>> More from SciFinder ^®
https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsbg%253D&md5=5ec23dce64722830afa856002acd6cf6

Cited By

ARTICLE SECTIONS

Jump To

This article is cited by 1324 publications.

Rupeng Dai, Xueting Bao, Ying Zhang, Yan Huang, Haohao Zhu, Kundi Yang, Bo Wang, Hongmei Wen, Wei Li, Jian Liu. Hot-Spot Residue-Based Virtual Screening of Novel Selective Estrogen-Receptor Degraders for Breast Cancer Treatment. Journal of Chemical Information and Modeling 2023, 63 (23) , 7588-7602. https://doi.org/10.1021/acs.jcim.3c01503
Luise Jacobsen, Jonathan Hungerland, Vladimir Bačić, Luca Gerhards, Fabian Schuhmann, Ilia A. Solov’yov. Introducing the Automated Ligand Searcher. Journal of Chemical Information and Modeling 2023, 63 (23) , 7518-7528. https://doi.org/10.1021/acs.jcim.3c01317
Daniel Del Hoyo, Martin Salinas, Alba Lomas, Eugenia Ulzurrun, Nuria E. Campillo, Carlos Oscar Sorzano. Scipion-Chem: An Open Platform for Virtual Drug Screening. Journal of Chemical Information and Modeling 2023, Article ASAP.
Mehdi Paykan Heyrati, Zahra Ghorbanali, Mohammad Akbari, Ghasem Pishgahi, Fatemeh Zare-Mirakabad. BioAct-Het: A Heterogeneous Siamese Neural Network for Bioactivity Prediction Using Novel Bioactivity Representation. ACS Omega 2023, 8 (47) , 44757-44772. https://doi.org/10.1021/acsomega.3c05778
Furyal Ahmed, Charles L. Brooks, III. FASTDock: A Pipeline for Allosteric Drug Discovery. Journal of Chemical Information and Modeling 2023, 63 (22) , 7219-7227. https://doi.org/10.1021/acs.jcim.3c00895
Manuel A. Llanos, Nicolás Enrique, Vega Esteban-López, Sebastian Scioli-Montoto, David Sánchez-Benito, María E. Ruiz, Veronica Milesi, Dolores E. López, Alan Talevi, Pedro Martín, Luciana Gavernet. A Combined Ligand- and Structure-Based Virtual Screening To Identify Novel NaV1.2 Blockers: In Vitro Patch Clamp Validation and In Vivo Anticonvulsant Activity. Journal of Chemical Information and Modeling 2023, 63 (22) , 7083-7096. https://doi.org/10.1021/acs.jcim.3c00645
Andrew T. McNutt, Fatimah Bisiriyu, Sophia Song, Ananya Vyas, Geoffrey R. Hutchison, David Ryan Koes. Conformer Generation for Structure-Based Drug Design: How Many and How Good?. Journal of Chemical Information and Modeling 2023, 63 (21) , 6598-6607. https://doi.org/10.1021/acs.jcim.3c01245
Zixuan Cheng, Siaw San Hwang, Mrinal Bhave, Taufiq Rahman, Xavier Chee Wezen. Combination of QSAR Modeling and Hybrid-Based Consensus Scoring to Identify Dual-Targeting Inhibitors of PLK1 and p38γ. Journal of Chemical Information and Modeling 2023, 63 (21) , 6912-6924. https://doi.org/10.1021/acs.jcim.3c01252
Lan Phuong Nguyen, Rasel Ahmed Khan, Soomin Kang, Hobin Lee, Jong-Ik Hwang, Hong-Rae Kim. Discovery of Chemical Scaffolds as Lysophosphatidic Acid Receptor 1 Antagonists: Virtual Screening, In Vitro Validation, and Molecular Dynamics Analysis. ACS Omega 2023, 8 (43) , 40375-40386. https://doi.org/10.1021/acsomega.3c04798
Mohemmed Faraz Khan, Shubhangi Kandwal, Darren Fayne. DataPype: A Fully Automated Unified Software Platform for Computer-Aided Drug Design. ACS Omega 2023, 8 (42) , 39468-39480. https://doi.org/10.1021/acsomega.3c05207
Muhammad Yasir, Jinyoung Park, Eun-Taek Han, Won Sun Park, Jin-Hee Han, Yong-Soo Kwon, Hee-Jae Lee, Wanjoo Chun. Vismodegib Identified as a Novel COX-2 Inhibitor via Deep-Learning-Based Drug Repositioning and Molecular Docking Analysis. ACS Omega 2023, 8 (37) , 34160-34170. https://doi.org/10.1021/acsomega.3c05425
Monica A. Kamal, Hedy A. Badary, Dalia Omran, Hend I. Shousha, Ashraf O. Abdelaziz, Hend M. El Tayebi, Yasmine M. Mandour. Virtual Screening and Biological Evaluation of Potential PD-1/PD-L1 Immune Checkpoint Inhibitors as Anti-Hepatocellular Carcinoma Agents. ACS Omega 2023, 8 (37) , 33242-33254. https://doi.org/10.1021/acsomega.3c00279
Yan Li, Zhe Zhang, Renxiao Wang. HydraMap v.2: Prediction of Hydration Sites and Desolvation Energy with Refined Statistical Potentials. Journal of Chemical Information and Modeling 2023, 63 (15) , 4749-4761. https://doi.org/10.1021/acs.jcim.3c00408
Xujun Zhang, Chao Shen, Tianyue Wang, Yu Kang, Dan Li, Peichen Pan, Jike Wang, Gaoang Wang, Yafeng Deng, Lei Xu, Dongsheng Cao, Tingjun Hou, Zhe Wang. Topology-Based and Conformation-Based Decoys Database: An Unbiased Online Database for Training and Benchmarking Machine-Learning Scoring Functions. Journal of Medicinal Chemistry 2023, 66 (13) , 9174-9183. https://doi.org/10.1021/acs.jmedchem.3c00801
Xiaoyang Qu, Lina Dong, Ding Luo, Yubing Si, Binju Wang. Water Network-Augmented Two-State Model for Protein–Ligand Binding Affinity Prediction. Journal of Chemical Information and Modeling 2023, Article ASAP.
Song Li, Chao Hu, Song Ke, Chenxing Yang, Jun Chen, Yi Xiong, Hao Liu, Liang Hong. LS-MolGen: Ligand-and-Structure Dual-Driven Deep Reinforcement Learning for Target-Specific Molecular Generation Improves Binding Affinity and Novelty. Journal of Chemical Information and Modeling 2023, 63 (13) , 4207-4215. https://doi.org/10.1021/acs.jcim.3c00587
Isha Singh, Fengling Li, Elissa A. Fink, Irene Chau, Alice Li, Annía Rodriguez-Hernández, Isabella Glenn, Francisco J. Zapatero-Belinchón, M. Luis Rodriguez, Kanchan Devkota, Zhijie Deng, Kris White, Xiaobo Wan, Nataliya A. Tolmachova, Yurii S. Moroz, H. Ümit Kaniskan, Melanie Ott, Adolfo García-Sastre, Jian Jin, Danica Galonić Fujimori, John J. Irwin, Masoud Vedadi, Brian K. Shoichet. Structure-Based Discovery of Inhibitors of the SARS-CoV-2 Nsp14 N7-Methyltransferase. Journal of Medicinal Chemistry 2023, 66 (12) , 7785-7803. https://doi.org/10.1021/acs.jmedchem.2c02120
Xiangying Zhang, Haotian Gao, Haojie Wang, Zhihang Chen, Zhe Zhang, Xinchong Chen, Yan Li, Yifei Qi, Renxiao Wang. PLANET: A Multi-objective Graph Neural Network Model for Protein–Ligand Binding Affinity Prediction. Journal of Chemical Information and Modeling 2023, Article ASAP.
Yuejiang Yu, Chun Cai, Jiayue Wang, Zonghua Bo, Zhengdan Zhu, Hang Zheng. Uni-Dock: GPU-Accelerated Docking Enables Ultralarge Virtual Screening. Journal of Chemical Theory and Computation 2023, 19 (11) , 3336-3345. https://doi.org/10.1021/acs.jctc.2c01145
Xu Qian, Xiaowen Dai, Lin Luo, Mingde Lin, Yuan Xu, Yang Zhao, Dingfang Huang, Haodi Qiu, Li Liang, Haichun Liu, Yingbo Liu, Lingxi Gu, Tao Lu, Yadong Chen, Yanmin Zhang. An Interpretable Multitask Framework BiLAT Enables Accurate Prediction of Cyclin-Dependent Protein Kinase Inhibitors. Journal of Chemical Information and Modeling 2023, 63 (11) , 3350-3368. https://doi.org/10.1021/acs.jcim.3c00473
Yuwei Yang, Chang-Yu Hsieh, Yu Kang, Tingjun Hou, Huanxiang Liu, Xiaojun Yao. Deep Generation Model Guided by the Docking Score for Active Molecular Design. Journal of Chemical Information and Modeling 2023, 63 (10) , 2983-2991. https://doi.org/10.1021/acs.jcim.3c00572
Jerome Eberhardt, Stefano Forli. WaterKit: Thermodynamic Profiling of Protein Hydration Sites. Journal of Chemical Theory and Computation 2023, 19 (9) , 2535-2556. https://doi.org/10.1021/acs.jctc.2c01087
Christian Kersten, Steven Clower, Fabian Barthels. Hic Sunt Dracones: Molecular Docking in Uncharted Territories with Structures from AlphaFold2 and RoseTTAfold. Journal of Chemical Information and Modeling 2023, 63 (7) , 2218-2225. https://doi.org/10.1021/acs.jcim.2c01400
Anna M. Díaz-Rovira, Helena Martín, Thijs Beuming, Lucía Díaz, Victor Guallar, Soumya S. Ray. Are Deep Learning Structural Models Sufficiently Accurate for Virtual Screening? Application of Docking Algorithms to AlphaFold2 Predicted Structures. Journal of Chemical Information and Modeling 2023, 63 (6) , 1668-1674. https://doi.org/10.1021/acs.jcim.2c01270
Yuqi Zhang, Marton Vass, Da Shi, Esam Abualrous, Jennifer M. Chambers, Nikita Chopra, Christopher Higgs, Koushik Kasavajhala, Hubert Li, Prajwal Nandekar, Hideyuki Sato, Edward B. Miller, Matthew P. Repasky, Steven V. Jerome. Benchmarking Refined and Unrefined AlphaFold2 Structures for Hit Discovery. Journal of Chemical Information and Modeling 2023, 63 (6) , 1656-1667. https://doi.org/10.1021/acs.jcim.2c01219
Yeajee Kwon, Sera Park, Jaeok Lee, Jiyeon Kang, Hwa Jeong Lee, Wankyu Kim. BEAR: A Novel Virtual Screening Method Based on Large-Scale Bioactivity Data. Journal of Chemical Information and Modeling 2023, 63 (5) , 1429-1437. https://doi.org/10.1021/acs.jcim.2c01300
Lukas Waterloo, Harald Hübner, Fabrizio Fierro, Tara Pfeiffer, Regine Brox, Stefan Löber, Dorothee Weikert, Masha Y. Niv, Peter Gmeiner. Discovery of 2-Aminopyrimidines as Potent Agonists for the Bitter Taste Receptor TAS2R14. Journal of Medicinal Chemistry 2023, 66 (5) , 3499-3521. https://doi.org/10.1021/acs.jmedchem.2c01997
Izaz Monir Kamal, Saikat Chakrabarti. MetaDOCK: A Combinatorial Molecular Docking Approach. ACS Omega 2023, 8 (6) , 5850-5860. https://doi.org/10.1021/acsomega.2c07619
Horrick Sharma, Pragya Sharma, Uzziah Urquiza, Lerin R. Chastain, Michael A. Ihnat. Exploration of a Large Virtual Chemical Space: Identification of Potent Inhibitors of Lactate Dehydrogenase-A against Pancreatic Cancer. Journal of Chemical Information and Modeling 2023, 63 (3) , 1028-1043. https://doi.org/10.1021/acs.jcim.2c01544
Piseth Nhoek, Sungjin Ahn, Pisey Pel, Young-Mi Kim, Jungmoo Huh, Hyun Woo Kim, Minsoo Noh, Young-Won Chin. Alkaloids and Coumarins with Adiponectin-Secretion-Promoting Activities from the Leaves of Orixa japonica. Journal of Natural Products 2023, 86 (1) , 138-148. https://doi.org/10.1021/acs.jnatprod.2c00844
Ganesh Chandan Kanakala, Rishal Aggarwal, Divya Nayar, U. Deva Priyakumar. Latent Biases in Machine Learning Models for Predicting Binding Affinities Using Popular Data Sets. ACS Omega 2023, 8 (2) , 2389-2397. https://doi.org/10.1021/acsomega.2c06781
Jörg Heider, Jonas Kilian, Aleksandra Garifulina, Steffen Hering, Thierry Langer, Thomas Seidel. Apo2ph4: A Versatile Workflow for the Generation of Receptor-based Pharmacophore Models for Virtual Screening. Journal of Chemical Information and Modeling 2023, 63 (1) , 101-110. https://doi.org/10.1021/acs.jcim.2c00814
Daniel Vella, Jean-Paul Ebejer. Few-Shot Learning for Low-Data Drug Discovery. Journal of Chemical Information and Modeling 2023, 63 (1) , 27-42. https://doi.org/10.1021/acs.jcim.2c00779
Fergus Boyles, Charlotte M. Deane, Garrett M. Morris. Learning from Docked Ligands: Ligand-Based Features Rescue Structure-Based Scoring Functions When Trained on Docked Poses. Journal of Chemical Information and Modeling 2022, 62 (22) , 5329-5341. https://doi.org/10.1021/acs.jcim.1c00096
Eric R. Hantz, Steffen Lindert. Actives-Based Receptor Selection Strongly Increases the Success Rate in Structure-Based Drug Design and Leads to Identification of 22 Potent Cancer Inhibitors. Journal of Chemical Information and Modeling 2022, 62 (22) , 5675-5687. https://doi.org/10.1021/acs.jcim.2c00848
Connor J. Morris, Jacob A. Stern, Brenden Stark, Max Christopherson, Dennis Della Corte. MILCDock: Machine Learning Enhanced Consensus Docking for Virtual Screening in Drug Discovery. Journal of Chemical Information and Modeling 2022, 62 (22) , 5342-5350. https://doi.org/10.1021/acs.jcim.2c00705
Janez Konc, Dušanka Janežič. ProBiS-Fold Approach for Annotation of Human Structures from the AlphaFold Database with No Corresponding Structure in the PDB to Discover New Druggable Binding Sites. Journal of Chemical Information and Modeling 2022, 62 (22) , 5821-5829. https://doi.org/10.1021/acs.jcim.2c00947
Yanjun Li, Daohong Zhou, Guangrong Zheng, Xiaolin Li, Dapeng Wu, Yaxia Yuan. DyScore: A Boosting Scoring Method with Dynamic Properties for Identifying True Binders and Nonbinders in Structure-Based Drug Discovery. Journal of Chemical Information and Modeling 2022, 62 (22) , 5550-5567. https://doi.org/10.1021/acs.jcim.2c00926
Jiao Zhou, Wei Li, Shanyue Guan, Xiaohong Chen, Xiang Liu, Weiyan Shao. Discovery of Chemokine CXCL12 Inhibitors by Tandem Application of Virtual Screening and NMR Spectrometry. Journal of Chemical Information and Modeling 2022, 62 (22) , 5729-5737. https://doi.org/10.1021/acs.jcim.2c01018
Jinze Zhang, Hao Li, Xuejun Zhao, Qilong Wu, Sheng-You Huang. Holo Protein Conformation Generation from Apo Structures by Ligand Binding Site Refinement. Journal of Chemical Information and Modeling 2022, 62 (22) , 5806-5820. https://doi.org/10.1021/acs.jcim.2c00895
Min Xu, Cheng Shen, Jincai Yang, Qing Wang, Niu Huang. Systematic Investigation of Docking Failures in Large-Scale Structure-Based Virtual Screening. ACS Omega 2022, 7 (43) , 39417-39428. https://doi.org/10.1021/acsomega.2c05826
Juliana María García-Chacón, Edisson Tello, Ericsson Coy-Barrera, Devin G. Peterson, Coralia Osorio. Mono-n-butyl Malate-Derived Compounds from Camu-camu (Myrciaria dubia) Malic Acid: The Alkyl-Dependent Antihyperglycemic-Related Activity. ACS Omega 2022, 7 (43) , 39335-39346. https://doi.org/10.1021/acsomega.2c05551
Timothy R. Stachowski, Marcus Fischer. Large-Scale Ligand Perturbations of the Protein Conformational Landscape Reveal State-Specific Interaction Hotspots. Journal of Medicinal Chemistry 2022, 65 (20) , 13692-13704. https://doi.org/10.1021/acs.jmedchem.2c00708
Stefanie Kampen, David Rodríguez, Morten Jørgensen, Monika Kruszyk-Kujawa, Xinyan Huang, Michael Collins, Jr, Noel Boyle, Damien Maurel, Axel Rudling, Guillaume Lebon, Jens Carlsson. Structure-Based Discovery of Negative Allosteric Modulators of the Metabotropic Glutamate Receptor 5. ACS Chemical Biology 2022, 17 (10) , 2744-2752. https://doi.org/10.1021/acschembio.2c00234
Adam Stasiulewicz, Anna Lesniak, Piotr Setny, Magdalena Bujalska-Zadrożny, Joanna I. Sulkowska. Identification of CB1 Ligands among Drugs, Phytochemicals and Natural-Like Compounds: Virtual Screening and In Vitro Verification. ACS Chemical Neuroscience 2022, 13 (20) , 2991-3007. https://doi.org/10.1021/acschemneuro.2c00502
Agamemnon Krasoulis, Nick Antonopoulos, Vassilis Pitsikalis, Stavros Theodorakis. DENVIS: Scalable and High-Throughput Virtual Screening Using Graph Neural Networks with Atomic and Surface Protein Pocket Features. Journal of Chemical Information and Modeling 2022, 62 (19) , 4642-4659. https://doi.org/10.1021/acs.jcim.2c01057
Melisa E. Gantner, Denis N. Prada Gori, Manuel A. Llanos, Alan Talevi, Andrea Angeli, Daniela Vullo, Claudiu T. Supuran, Luciana Gavernet. Identification of New Carbonic Anhydrase VII Inhibitors by Structure-Based Virtual Screening. Journal of Chemical Information and Modeling 2022, 62 (19) , 4760-4770. https://doi.org/10.1021/acs.jcim.2c00910
Baddipadige Raju, Gera Narendra, Himanshu Verma, Manoj Kumar, Bharti Sapra, Gurleen Kaur, Subheet Kumar jain, Om Silakari. Machine Learning Enabled Structure-Based Drug Repurposing Approach to Identify Potential CYP1B1 Inhibitors. ACS Omega 2022, 7 (36) , 31999-32013. https://doi.org/10.1021/acsomega.2c02983
Elisabeth Kallert, Tim R. Fischer, Simon Schneider, Maike Grimm, Mark Helm, Christian Kersten. Protein-Based Virtual Screening Tools Applied for RNA–Ligand Docking Identify New Binders of the preQ1-Riboswitch. Journal of Chemical Information and Modeling 2022, 62 (17) , 4134-4148. https://doi.org/10.1021/acs.jcim.2c00751
Keisuke Yanagisawa, Rikuto Kubota, Yasushi Yoshikawa, Masahito Ohue, Yutaka Akiyama. Effective Protein–Ligand Docking Strategy via Fragment Reuse and a Proof-of-Concept Implementation. ACS Omega 2022, 7 (34) , 30265-30274. https://doi.org/10.1021/acsomega.2c03470
Chao Shen, Xujun Zhang, Yafeng Deng, Junbo Gao, Dong Wang, Lei Xu, Peichen Pan, Tingjun Hou, Yu Kang. Boosting Protein–Ligand Binding Pose Prediction and Virtual Screening Based on Residue–Atom Distance Likelihood Potential and Graph Transformer. Journal of Medicinal Chemistry 2022, 65 (15) , 10691-10706. https://doi.org/10.1021/acs.jmedchem.2c00991
Miguel García-Ortegón, Gregor N. C. Simm, Austin J. Tripp, José Miguel Hernández-Lobato, Andreas Bender, Sergio Bacallado. DOCKSTRING: Easy Molecular Docking Yields Better Benchmarks for Ligand Design. Journal of Chemical Information and Modeling 2022, 62 (15) , 3486-3502. https://doi.org/10.1021/acs.jcim.1c01334
Manuel A. Llanos, Nicolás Enrique, María L. Sbaraglini, Federico M. Garofalo, Alan Talevi, Luciana Gavernet, Pedro Martín. Structure-Based Virtual Screening Identifies Novobiocin, Montelukast, and Cinnarizine as TRPV1 Modulators with Anticonvulsant Activity In Vivo. Journal of Chemical Information and Modeling 2022, 62 (12) , 3008-3022. https://doi.org/10.1021/acs.jcim.2c00312
Haoxi Li, Rosa Mirabel, Joseph Zimmerman, Ion Ghiviriga, Darian K. Phidd, Nicole Horenstein, Nikhil M. Urs. Structure–Functional Selectivity Relationship Studies on A-86929 Analogs and Small Aryl Fragments toward the Discovery of Biased Dopamine D1 Receptor Agonists. ACS Chemical Neuroscience 2022, 13 (12) , 1818-1831. https://doi.org/10.1021/acschemneuro.2c00235
Michael C. Hutter. Differential Multimolecule Fingerprint for Similarity Search─Making Use of Active and Inactive Compound Sets in Virtual Screening. Journal of Chemical Information and Modeling 2022, 62 (11) , 2726-2736. https://doi.org/10.1021/acs.jcim.2c00242
Chao Yang, Yingkai Zhang. Delta Machine Learning to Improve Scoring-Ranking-Screening Performances of Protein–Ligand Scoring Functions. Journal of Chemical Information and Modeling 2022, 62 (11) , 2696-2712. https://doi.org/10.1021/acs.jcim.2c00485
Xujun Zhang, Chao Shen, Ben Liao, Dejun Jiang, Jike Wang, Zhenxing Wu, Hongyan Du, Tianyue Wang, Wenbo Huo, Lei Xu, Dongsheng Cao, Chang-Yu Hsieh, Tingjun Hou. TocoDecoy: A New Approach to Design Unbiased Datasets for Training and Benchmarking Machine-Learning Scoring Functions. Journal of Medicinal Chemistry 2022, 65 (11) , 7918-7932. https://doi.org/10.1021/acs.jmedchem.2c00460
Weixin Xie, Fanhao Wang, Yibo Li, Luhua Lai, Jianfeng Pei. Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models. Journal of Chemical Information and Modeling 2022, 62 (10) , 2269-2279. https://doi.org/10.1021/acs.jcim.2c00042
Nemanja Djokovic, Dusan Ruzic, Minna Rahnasto-Rilla, Tatjana Srdic-Rajic, Maija Lahtela-Kakkonen, Katarina Nikolic. Expanding the Accessible Chemical Space of SIRT2 Inhibitors through Exploration of Binding Pocket Dynamics. Journal of Chemical Information and Modeling 2022, 62 (10) , 2571-2585. https://doi.org/10.1021/acs.jcim.2c00241
Haoqi Wang, Nirmitee Mulgaonkar, Lisa M. Pérez, Sandun Fernando. ELIXIR-A: An Interactive Visualization Tool for Multi-Target Pharmacophore Refinement. ACS Omega 2022, 7 (15) , 12707-12715. https://doi.org/10.1021/acsomega.1c07144
Nabeel Ahmad, Anamika Singh, Akshita Gupta, Pradeep Pant, Tej P. Singh, Sujata Sharma, Pradeep Sharma. Discovery of the Lead Molecules Targeting the First Step of the Histidine Biosynthesis Pathway of Acinetobacter baumannii. Journal of Chemical Information and Modeling 2022, 62 (7) , 1744-1759. https://doi.org/10.1021/acs.jcim.1c01421
Tomomi Shimazaki, Masanori Tachikawa. Collaborative Approach between Explainable Artificial Intelligence and Simplified Chemical Interactions to Explore Active Ligands for Cyclin-Dependent Kinase 2. ACS Omega 2022, 7 (12) , 10372-10381. https://doi.org/10.1021/acsomega.1c06976
C. Johan van der Westhuizen, André Stander, Darren L. Riley, Jenny-Lee Panayides. Discovery of Novel Acetylcholinesterase Inhibitors by Virtual Screening, In Vitro Screening, and Molecular Dynamics Simulations. Journal of Chemical Information and Modeling 2022, 62 (6) , 1550-1572. https://doi.org/10.1021/acs.jcim.1c01443
Janez Konc, Samo Lešnik, Blaž Škrlj, Matej Sova, Matic Proj, Damijan Knez, Stanislav Gobec, Dušanka Janežič. ProBiS-Dock: A Hybrid Multitemplate Homology Flexible Docking Algorithm Enabled by Protein Binding Site Comparison. Journal of Chemical Information and Modeling 2022, 62 (6) , 1573-1584. https://doi.org/10.1021/acs.jcim.1c01176
Giovanni Bolcato, Esther Heid, Jonas Boström. On the Value of Using 3D Shape and Electrostatic Similarities in Deep Generative Methods. Journal of Chemical Information and Modeling 2022, 62 (6) , 1388-1398. https://doi.org/10.1021/acs.jcim.1c01535
Anat Levit Kaplan, Ryan T. Strachan, Joao M. Braz, Veronica Craik, Samuel Slocum, Thomas Mangano, Vanessa Amabo, Henry O’Donnell, Parnian Lak, Allan I. Basbaum, Bryan L. Roth, Brian K. Shoichet. Structure-Based Design of a Chemical Probe Set for the 5-HT5A Serotonin Receptor. Journal of Medicinal Chemistry 2022, 65 (5) , 4201-4217. https://doi.org/10.1021/acs.jmedchem.1c02031
Eugene Lin, Chieh-Hsin Lin, Hsien-Yuan Lane. De Novo Peptide and Protein Design Using Generative Adversarial Networks: An Update. Journal of Chemical Information and Modeling 2022, 62 (4) , 761-774. https://doi.org/10.1021/acs.jcim.1c01361
Sami T. Kurkinen, Jukka V. Lehtonen, Olli T. Pentikäinen, Pekka A. Postila. Optimization of Cavity-Based Negative Images to Boost Docking Enrichment in Virtual Screening. Journal of Chemical Information and Modeling 2022, 62 (4) , 1100-1112. https://doi.org/10.1021/acs.jcim.1c01145
Fabio Begnini, Stefan Geschwindner, Patrik Johansson, Lisa Wissler, Richard J. Lewis, Emma Danelius, Andreas Luttens, Pierre Matricon, Jens Carlsson, Stijn Lenders, Beate König, Anna Friedel, Peter Sjö, Stefan Schiesser, Jan Kihlberg. Importance of Binding Site Hydration and Flexibility Revealed When Optimizing a Macrocyclic Inhibitor of the Keap1–Nrf2 Protein–Protein Interaction. Journal of Medicinal Chemistry 2022, 65 (4) , 3473-3517. https://doi.org/10.1021/acs.jmedchem.1c01975
Andreas Luttens, Hjalmar Gullberg, Eldar Abdurakhmanov, Duy Duc Vo, Dario Akaberi, Vladimir O. Talibov, Natalia Nekhotiaeva, Laura Vangeel, Steven De Jonghe, Dirk Jochmans, Janina Krambrich, Ali Tas, Bo Lundgren, Ylva Gravenfors, Alexander J. Craig, Yoseph Atilaw, Anja Sandström, Lindon W. K. Moodie, Åke Lundkvist, Martijn J. van Hemert, Johan Neyts, Johan Lennerstrand, Jan Kihlberg, Kristian Sandberg, U. Helena Danielson, Jens Carlsson. Ultralarge Virtual Screening Identifies SARS-CoV-2 Main Protease Inhibitors with Broad-Spectrum Activity against Coronaviruses. Journal of the American Chemical Society 2022, 144 (7) , 2905-2920. https://doi.org/10.1021/jacs.1c08402
Dongping Li, Kexin Jiang, Dan Teng, Zengrui Wu, Weihua Li, Yun Tang, Rui Wang, Guixia Liu. Discovery of New Estrogen-Related Receptor α Agonists via a Combination Strategy Based on Shape Screening and Ensemble Docking. Journal of Chemical Information and Modeling 2022, 62 (3) , 486-497. https://doi.org/10.1021/acs.jcim.1c00662
Damien Geslin, Alban Lepailleur, Jean-Luc Manguin, Nhat-Vinh Vo, Jean-Luc Lamotte, Bertrand Cuissart, Ronan Bureau. Deciphering a Pharmacophore Network: A Case Study Using BCR-ABL Data. Journal of Chemical Information and Modeling 2022, 62 (3) , 678-691. https://doi.org/10.1021/acs.jcim.1c00427
Wenyi Zhang, Jing Huang. EViS: An Enhanced Virtual Screening Approach Based on Pocket–Ligand Similarity. Journal of Chemical Information and Modeling 2022, 62 (3) , 498-510. https://doi.org/10.1021/acs.jcim.1c00944
Iván Felsztyna, Marcos A. Villarreal, Daniel A. García, Virginia Miguel. Insect RDL Receptor Models for Virtual Screening: Impact of the Template Conformational State in Pentameric Ligand-Gated Ion Channels. ACS Omega 2022, 7 (2) , 1988-2001. https://doi.org/10.1021/acsomega.1c05465
Ilenia Giangreco, Abhik Mukhopadhyay, Jason C. Cole. Validation of a Field-Based Ligand Screener Using a Novel Benchmarking Data Set for Assessing 3D-Based Virtual Screening Methods. Journal of Chemical Information and Modeling 2021, 61 (12) , 5841-5852. https://doi.org/10.1021/acs.jcim.1c00866
Dejun Jiang, Chang-Yu Hsieh, Zhenxing Wu, Yu Kang, Jike Wang, Ercheng Wang, Ben Liao, Chao Shen, Lei Xu, Jian Wu, Dongsheng Cao, Tingjun Hou. InteractionGraphNet: A Novel and Efficient Deep Graph Representation Learning Framework for Accurate Protein–Ligand Interaction Predictions. Journal of Medicinal Chemistry 2021, 64 (24) , 18209-18232. https://doi.org/10.1021/acs.jmedchem.1c01830
Yujin Wu, Charles L. Brooks III. Flexible CDOCKER: Hybrid Searching Algorithm and Scoring Function with Side Chain Conformational Entropy. Journal of Chemical Information and Modeling 2021, 61 (11) , 5535-5549. https://doi.org/10.1021/acs.jcim.1c01078
Hugo Guterres, Sang-Jun Park, Yiwei Cao, Wonpil Im. CHARMM-GUI Ligand Designer for Template-Based Virtual Ligand Design in a Binding Site. Journal of Chemical Information and Modeling 2021, 61 (11) , 5336-5342. https://doi.org/10.1021/acs.jcim.1c01156
Anantha Krishnan Dhanabalan, Mamangam Subaraja, Kuppusamy Palanichamy, Devadasan Velmurugan, Krishnasamy Gunasekaran. Identification of a Chlorogenic Ester as a Monoamine Oxidase (MAO-B) Inhibitor by Integrating “Traditional and Machine Learning” Virtual Screening and In Vitro as well as In Vivo Validation: A Lead against Neurodegenerative Disorders?. ACS Chemical Neuroscience 2021, 12 (19) , 3690-3707. https://doi.org/10.1021/acschemneuro.1c00430
Shuo Gu, Matthew S. Smith, Ying Yang, John J. Irwin, Brian K. Shoichet. Ligand Strain Energy in Large Library Docking. Journal of Chemical Information and Modeling 2021, 61 (9) , 4331-4341. https://doi.org/10.1021/acs.jcim.1c00368
Panagiotis I. Koukos, Manon Réau, Alexandre M. J. J. Bonvin. Shape-Restrained Modeling of Protein–Small-Molecule Complexes with High Ambiguity Driven DOCKing. Journal of Chemical Information and Modeling 2021, 61 (9) , 4807-4818. https://doi.org/10.1021/acs.jcim.1c00796
Shuoyan Tan, Xiaoqing Gong, Huanxiang Liu, Xiaojun Yao. Virtual Screening and Biological Activity Evaluation of New Potent Inhibitors Targeting LRRK2 Kinase Domain. ACS Chemical Neuroscience 2021, 12 (17) , 3214-3224. https://doi.org/10.1021/acschemneuro.1c00399
Felix Musil, Andrea Grisafi, Albert P. Bartók, Christoph Ortner, Gábor Csányi, Michele Ceriotti. Physics-Inspired Structural Representations for Molecules and Materials. Chemical Reviews 2021, 121 (16) , 9759-9815. https://doi.org/10.1021/acs.chemrev.1c00021
Jerome Eberhardt, Diogo Santos-Martins, Andreas F. Tillack, Stefano Forli. AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. Journal of Chemical Information and Modeling 2021, 61 (8) , 3891-3898. https://doi.org/10.1021/acs.jcim.1c00203
Janez Konc, Samo Lešnik, Blaž Škrlj, Dušanka Janežič. ProBiS-Dock Database: A Web Server and Interactive Web Repository of Small Ligand–Protein Binding Sites for Drug Design. Journal of Chemical Information and Modeling 2021, 61 (8) , 4097-4107. https://doi.org/10.1021/acs.jcim.1c00454
Hugo Guterres, Sang-Jun Park, Han Zhang, Wonpil Im. CHARMM-GUI LBS Finder & Refiner for Ligand Binding Site Prediction and Refinement. Journal of Chemical Information and Modeling 2021, 61 (8) , 3744-3751. https://doi.org/10.1021/acs.jcim.1c00561
SahaIshikaGraduate Student Researcherishikasaha@g.ucla.eduHarranPatrick G.D.J. & J.M. Cram Chair in Organic Chemistryharran@chem.ucla.eduDr. Jonathan Bohmann, Department of Pharmaceuticals and Bioengineering, Southwest Research Institute, Ryan Gumpper, Postdoctoral Researcher, University of North Carolina at Chapel Hill. Virtual Screening for Chemists. 2021https://doi.org/10.1021/acsinfocus.7e5001
Biao Ma, Kei Terayama, Shigeyuki Matsumoto, Yuta Isaka, Yoko Sasakura, Hiroaki Iwata, Mitsugu Araki, Yasushi Okuno. Structure-Based de Novo Molecular Generator Combined with Artificial Intelligence and Docking Simulations. Journal of Chemical Information and Modeling 2021, 61 (7) , 3304-3313. https://doi.org/10.1021/acs.jcim.1c00679
Wei Chen, Guanxing Chen, Lu Zhao, Calvin Yu-Chian Chen. Predicting Drug–Target Interactions with Deep-Embedding Learning of Graphs and Sequences. The Journal of Physical Chemistry A 2021, 125 (25) , 5633-5642. https://doi.org/10.1021/acs.jpca.1c02419
Alzbeta Tuerkova, Orsolya Ungvári, Réka Laczkó-Rigó, Erzsébet Mernyák, Gergely Szakács, Csilla Özvegy-Laczka, Barbara Zdrazil. Data-Driven Ensemble Docking to Map Molecular Interactions of Steroid Analogs with Hepatic Organic Anion Transporting Polypeptides. Journal of Chemical Information and Modeling 2021, 61 (6) , 3109-3127. https://doi.org/10.1021/acs.jcim.1c00362
Lijuan Yang, Guanghui Yang, Xiaolong Chen, Qiong Yang, Xiaojun Yao, Zhitong Bing, Yuzhen Niu, Liang Huang, Lei Yang. Deep Scoring Neural Network Replacing the Scoring Function Components to Improve the Performance of Structure-Based Molecular Docking. ACS Chemical Neuroscience 2021, 12 (12) , 2133-2142. https://doi.org/10.1021/acschemneuro.1c00110
Raghuram Srinivas, Niraj Verma, Elfi Kraka, Eric C. Larson. Deep Learning-Based Ligand Design Using Shared Latent Implicit Fingerprints from Collaborative Filtering. Journal of Chemical Information and Modeling 2021, 61 (5) , 2159-2174. https://doi.org/10.1021/acs.jcim.0c01355
Francois Berenger, Ashutosh Kumar, Kam Y. J. Zhang, Yoshihiro Yamanishi. Lean-Docking: Exploiting Ligands’ Predicted Docking Scores to Accelerate Molecular Docking. Journal of Chemical Information and Modeling 2021, 61 (5) , 2341-2352. https://doi.org/10.1021/acs.jcim.0c01452
Hongyi Zhou, Hongnan Cao, Jeffrey Skolnick. FRAGSITE: A Fragment-Based Approach for Virtual Ligand Screening. Journal of Chemical Information and Modeling 2021, 61 (4) , 2074-2089. https://doi.org/10.1021/acs.jcim.0c01160
Joanna Zarnecka, Iva Lukac, Stephen J. Messham, Alhusein Hussin, Francesco Coppola, Steven J. Enoch, Alexander G. Dossetter, Edward J. Griffen, Andrew G. Leach. Mapping Ligand-Shape Space for Protein–Ligand Systems: Distinguishing Key-in-Lock and Hand-in-Glove Proteins. Journal of Chemical Information and Modeling 2021, 61 (4) , 1859-1874. https://doi.org/10.1021/acs.jcim.1c00089
Chao Li, Jun Sun, Vasile Palade. MSLDOCK: Multi-Swarm Optimization for Flexible Ligand Docking and Virtual Screening. Journal of Chemical Information and Modeling 2021, 61 (3) , 1500-1515. https://doi.org/10.1021/acs.jcim.0c01358
Reed M. Stein, Ying Yang, Trent E. Balius, Matt J. O’Meara, Jiankun Lyu, Jennifer Young, Khanh Tang, Brian K. Shoichet, John J. Irwin. Property-Unmatched Decoys in Docking Benchmarks. Journal of Chemical Information and Modeling 2021, 61 (2) , 699-714. https://doi.org/10.1021/acs.jcim.0c00598
Robert Schmidt, Florian Krull, Anna Lina Heinzke, Matthias Rarey. Disconnected Maximum Common Substructures under Constraints. Journal of Chemical Information and Modeling 2021, 61 (1) , 167-178. https://doi.org/10.1021/acs.jcim.0c00741
Katherine J. Schultz, Sean M. Colby, Vivian S. Lin, Aaron T. Wright, Ryan S. Renslow. Ligand- and Structure-Based Analysis of Deep Learning-Generated Potential α2a Adrenoceptor Agonists. Journal of Chemical Information and Modeling 2021, 61 (1) , 481-492. https://doi.org/10.1021/acs.jcim.0c01019
Hugo Guterres, Sang-Jun Park, Wei Jiang, Wonpil Im. Ligand-Binding-Site Refinement to Generate Reliable Holo Protein Structure Conformations from Apo Structures. Journal of Chemical Information and Modeling 2021, 61 (1) , 535-546. https://doi.org/10.1021/acs.jcim.0c01354

Load more citations

Download PDF

Abstract
High Resolution Image
Download MS PowerPoint Slide
Figure 1
Figure 1. DUD-E target classification. Number of the 102 targets that belong to eight broad protein categories.
High Resolution Image
Download MS PowerPoint Slide
Figure 2
Figure 2. Ligand clustering. (A) The seventh largest Murcko cluster of kinesin-like protein 1 (KIF11), showing both the scaffold (left) and all seven member ligands. (B) Number of ligands in each of the 70 KIF11 Bemis–Murcko atomic frameworks. We removed lower affinity compounds over-represented clusters (above the line), while retaining 100 ligands. (C) Number of adenosine A_2A receptor (AA2AR) Murcko clusters is plotted against affinity threshold. Fewer than 600 clusters are present using a 30 nM affinity threshold.
High Resolution Image
Download MS PowerPoint Slide
Figure 3
Figure 3. Decoy generation. (A) Three key “warhead” groups from factor Xa (FA10), glycinamide ribonucleotide transformylase (PUR2), and thymidine kinase (KITH). (B) Fraction of warheads remaining is plotted against the dissimilarity method. The dissimilarity methods consist of a fingerprint (Daylight or ECFP4) and either a hard cutoff or a fraction of the most dissimilar decoys to be retained. (C) Property distributions of estrogen receptor α (ESR1) for both the 383 ligands (blue) and the 20685 property-matched decoys (red).
High Resolution Image
Download MS PowerPoint Slide
Figure 4
Figure 4. Retrospective enrichment comparing ligand desolvation and electrostatics methods. Docking results over DUD-E as measured by LogAUC. “None” has no ligand desolvation term, “SEV” uses solvent-excluded volume ligand desolvation, “Thin” employs a thin low-dielectric layer in the electrostatic calculations.
High Resolution Image
Download MS PowerPoint Slide
Figure 5
Figure 5. Representative ROC plots. ROC plots using no desolvation (None), solvent-excluded volume ligand desolvation (SEV), the thin low-dielectric layer (Thin), or a drug-like background that consists of all ChEMBL12 ligands with affinities better than 10 μM (Drug-like). The black dotted line represents the results expected from docking ligands randomly. LogAUC percentages are reported in the legend text.
High Resolution Image
Download MS PowerPoint Slide
Figure 6
Figure 6. Representative docking poses. The crystallographic ligand was rebuilt and docked from scratch. (A–F) The crystal pose (magenta) is compared to the resulting docked pose (green). In (C), more ligand conformations are generated and the redocked pose is also shown (tan). Key hydrogen bonds are shown by black dotted lines, and the partially transparent protein surface is colored by atom type.
High Resolution Image
Download MS PowerPoint Slide
References
ARTICLE SECTIONS
Jump To
This article references 53 other publications.
1. 1
  Kitchen, D. B.; Decornez, H.; Furr, J. R.; Bajorath, J. Docking and scoring in virtual screening for drug discovery: methods and applications Nature Rev. Drug Discovery 2004, 3, 935– 949
  Google Scholar
  1
  Docking and scoring in virtual screening for drug discovery: methods and applications
  Kitchen, Douglas B.; Decornez, Helene; Furr, John R.; Bajorath, Juergen
  Nature Reviews Drug Discovery (2004), 3 (11), 935-949CODEN: NRDDAG; ISSN:1474-1776. (Nature Publishing Group)
  A review. Computational approaches that 'dock' small mols. into the structures of macromol. targets and 'score' their potential complementarity to binding sites are widely used in hit identification and lead optimization. Indeed, there are now a no. of drugs whose development was heavily influenced by or based on structure-based design and screening strategies, such as HIV protease inhibitors. Nevertheless, there remain significant challenges in the application of these approaches, in particular in relation to current scoring schemes. Here, we review key concepts and specific features of small-mol.-protein docking methods, highlight selected applications and discuss recent advances that aim to address the acknowledged limitations of established approaches.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXptFemtrg%253D&md5=875a2b37a4299181509a1922b11dbd2f
2. 2
  Kolb, P.; Rosenbaum, D. M.; Irwin, J. J.; Fung, J. J.; Kobilka, B. K.; Shoichet, B. K. Structure-based discovery of beta(2)-adrenergic receptor ligands Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 6843– 6848
  Google Scholar
  2
  Structure-based discovery of β2-adrenergic receptor ligands
  Kolb, Peter; Rosenbaum, Daniel M.; Irwin, John J.; Fung, Juan Jose; Kobilka, Brian K.; Shoichet, Brian K.
  Proceedings of the National Academy of Sciences of the United States of America (2009), 106 (16), 6843-6848CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
  Aminergic G protein-coupled receptors (GPCRs) have been a major focus of pharmaceutical research for many years. Due partly to the lack of reliable receptor structures, drug discovery efforts have been largely ligand-based. The recently detd. X-ray structure of the β2-adrenergic receptor offers an opportunity to investigate the advantages and limitations inherent in a structure-based approach to ligand discovery against this and related GPCR targets. Approx. 1 million com. available, "lead-like" mols. were docked against the β2-adrenergic receptor structure. On testing of 25 high-ranking mols., 6 were active with binding affinities <4 μM, with the best mol. binding with a Ki of 9 nM (95% confidence interval 7-10 nM). Five of these mols. were inverse agonists. The high hit rate, the high affinity of the most potent mol., the discovery of unprecedented chemotypes among the new inhibitors, and the apparent bias toward inverse agonists among the docking hits, have implications for structure-based approaches against GPCRs that recognize small org. mols.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXlsV2qsro%253D&md5=4d1a4cb2aa3925aa4c99c6b0496417a7
3. 3
  Mysinger, M. M.; Weiss, D. R.; Ziarek, J. J.; Gravel, S.; Doak, A. K.; Karpiak, J.; Heveker, N.; Shoichet, B. K.; Volkman, B. F. Structure-based ligand discovery for the protein–protein interface of chemokine receptor CXCR4 Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 5517– 5522
  Google Scholar
  There is no corresponding record for this reference.
4. 4
  Gruneberg, S.; Stubbs, M. T.; Klebe, G. Successful virtual screening for novel inhibitors of human carbonic anhydrase: strategy and experimental confirmation J. Med. Chem. 2002, 45, 3588– 3602
  Google Scholar
  There is no corresponding record for this reference.
5. 5
  Jain, A. N.; Nicholls, A. Recommendations for evaluation of computational methods J. Comput.-Aided Mol. Des. 2008, 22, 133– 139
  Google Scholar
  5
  Recommendations for evaluation of computational methods
  Jain, Ajay N.; Nicholls, Anthony
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 133-139CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  A review. The field of computational chem., particularly as applied to drug design, has become increasingly important in terms of the practical application of predictive modeling to pharmaceutical research and development. Tools for exploiting protein structures or sets of ligands known to bind particular targets can be used for binding-mode prediction, virtual screening, and prediction of activity. A serious weakness within the field is a lack of stds. with respect to quant. evaluation of methods, data set prepn., and data set sharing. Our goal should be to report new methods or comparative evaluations of methods in a manner that supports decision making for practical applications. Here we propose a modest beginning, with recommendations for requirements on statistical reporting, requirements for data sharing, and best practices for benchmark prepn. and usage.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsb0%253D&md5=9576e150079dcbf6e4bfea5070082016
6. 6
  Babaoglu, K.; Simeonov, A.; Irwin, J. J.; Nelson, M. E.; Feng, B.; Thomas, C. J.; Cancian, L.; Costi, M. P.; Maltby, D. A.; Jadhav, A.; Inglese, J.; Austin, C. P.; Shoichet, B. K. Comprehensive mechanistic analysis of hits from high-throughput and docking screens against beta-lactamase J. Med. Chem. 2008, 51, 2502– 2511
  Google Scholar
  6
  Comprehensive Mechanistic Analysis of Hits from High-Throughput and Docking Screens against β-Lactamase
  Babaoglu, Kerim; Simeonov, Anton; Irwin, John J.; Nelson, Michael E.; Feng, Brian; Thomas, Craig J.; Cancian, Laura; Costi, M. Paola; Maltby, David A.; Jadhav, Ajit; Inglese, James; Austin, Christopher P.; Shoichet, Brian K.
  Journal of Medicinal Chemistry (2008), 51 (8), 2502-2511CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  High-throughput screening (HTS) is widely used in drug discovery. Esp. for screens of unbiased libraries, false positives can dominate "hit lists"; their origins are much debated. Here we det. the mechanism of every active hit from a screen of 70,563 unbiased mols. against β-lactamase using quant. HTS (qHTS). Of the 1274 initial inhibitors, 95% were detergent-sensitive and were classified as aggregators. Among the 70 remaining were 25 potent, covalent-acting β-lactams. Mass spectra, counter-screens, and crystallog. identified 12 as promiscuous covalent inhibitors. The remaining 33 were either aggregators or irreproducible. No specific reversible inhibitors were found. We turned to mol. docking to prioritize mols. from the same library for testing at higher concns. Of 16 tested, 2 were modest inhibitors. Subsequent X-ray structures corresponded to the docking prediction. Analog synthesis improved affinity to 8 μM. These results suggest that it may be the phys. behavior of org. mols., not their reactivity, that accounts for most screening artifacts. Structure-based methods may prioritize weak-but-novel chemotypes in unbiased library screens.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjtFSksb0%253D&md5=7736429aeafc6b17a5d0182e24df0dd4
7. 7
  Ferreira, R. S.; Simeonov, A.; Jadhav, A.; Eidam, O.; Mott, B. T.; Keiser, M. J.; McKerrow, J. H.; Maloney, D. J.; Irwin, J. J.; Shoichet, B. K. Complementarity between a docking and a high-throughput screen in discovering new cruzain inhibitors J. Med. Chem. 2010, 53, 4891– 4905
  Google Scholar
  7
  Complementarity Between a Docking and a High-Throughput Screen in Discovering New Cruzain Inhibitors
  Ferreira, Rafaela S.; Simeonov, Anton; Jadhav, Ajit; Eidam, Oliv; Mott, Bryan T.; Keiser, Michael J.; McKerrow, James H.; Maloney, David J.; Irwin, John J.; Shoichet, Brian K.
  Journal of Medicinal Chemistry (2010), 53 (13), 4891-4905CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Virtual and high-throughput screens (HTS) should have complementary strengths and weaknesses, but studies that prospectively and comprehensively compare them are rare. We undertook a parallel docking and HTS screen of 197861 compds. against cruzain, a thiol protease target for Chagas disease, looking for reversible, competitive inhibitors. On workup, 99% of the hits were eliminated as false positives, yielding 146 well-behaved, competitive ligands. These fell into five chemotypes: two were prioritized by scoring among the top 0.1% of the docking-ranked library, two were prioritized by behavior in the HTS and by clustering, and one chemotype was prioritized by both approaches. Detn. of an inhibitor/cruzain crystal structure and comparison of the high-scoring docking hits to expt. illuminated the origins of docking false-negatives and false-positives. Prioritizing mols. that are both predicted by docking and are HTS-active yields well-behaved mols., relatively unobscured by the false-positives to which both techniques are individually prone.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXntlWrurk%253D&md5=0661663ff0802bf981a1997eeca9fa05
8. 8
  Gohlke, H.; Klebe, G. Approaches to the description and prediction of the binding affinity of small-molecule ligands to macromolecular receptors Angew. Chem., Int. Ed. Engl. 2002, 41, 2644– 2676
  Google Scholar
  8
  Approaches to the description and prediction of the binding affinity of small-molecule ligands to macromolecular receptors
  Gohlke, Holger; Klebe, Gerhard
  Angewandte Chemie, International Edition (2002), 41 (15), 2644-2676CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH)
  A review. The influence of a xenobiotic compd. on an organism is usually summarized by the expression biol. activity. If a controlled, therapeutically relevant, and regulatory action is obsd. the compd. has potential as a drug, otherwise its toxicity on the biol. system is of interest. However, what do we understand by the biol. activity. In principle, the overall effect on an organism has to be considered. However, because of the complexity of the interrelated processes involved, as a simplification primarily the "main action" on the organism is taken into consideration. On the mol. level, biol. activity corresponds to the binding of a (lowmol. wt.) compd. to a macromol. receptor, usually a protein. Enzymic reactions or signal-transduction cascades are thereby influenced with respect to their function for the organism. We regard this binding as a process under equil. conditions; thus, binding can be described as an assocn. or dissocn. process. Accordingly, biol. activity is expressed as the affinity of both partners for each other, as a thermodn. equil. quantity. How well do we understand these terms and how well are they theor. predictable today. The holy grail of rational drug design is the prediction of the biol. activity of a compd. The processes involving ligand binding are extremely complicated, both ligand and protein are flexible mols., and the energy inventory between the bound and unbound states must be considered in aq. soln. How sophisticated and reliable are our exptl. approaches to obtaining the necessary insight. The present review summarizes our current understanding of the binding affinity of a small-mol. ligand to a protein. Both theor. and empirical approaches for predicting binding affinity, starting from the three-dimensional structure of a protein-ligand complex, will be described and compared. Exptl. methods, primarily microcalorimetry, will be discussed. As a perspective, our own knowledge-based approach towards affinity prediction and exptl. data on factorizing binding contributions to protein-ligand binding will be presented.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38Xmt1entbw%253D&md5=d2e5e2490fe8909e5899a1a89630223f
9. 9
  Enyedy, I. J.; Egan, W. J. Can we use docking and scoring for hit-to-lead optimization? J. Comput.-Aided Mol. Des. 2008, 22, 161– 168
  Google Scholar
  9
  Can we use docking and scoring for hit-to-lead optimization?
  Enyedy, Istvan J.; Egan, William J.
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 161-168CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  Docking and scoring is currently one of the tools used for hit finding and hit-to-lead optimization when structural information about the target is known. Docking scores have been found useful for optimizing ligand binding to reproduce exptl. obsd. binding modes. The question is, can docking and scoring be used reliably for hit-to-lead optimization To illustrate the challenges of scoring for hit-to-lead optimization, the relationship of docking scores with exptl. detd. IC50 values measured inhouse were tested. The influences of the particular target, crystal structure, and the precision of the scoring function on the ability to differentiate between actives and inactives were analyzed by calcg. the area under the curve of receiver operator characteristic curves for docking scores. It was found that for the test sets considered, MW and sometimes ClogP were as useful as GlideScores and no significant difference was obsd. between SP and XP scores for differentiating between actives and inactives. Interpretation by an expert is still required to successfully utilize docking and scoring in hit-to-lead optimization.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsbk%253D&md5=24adfd0e8bb71c0e7b32a9208e7ab4d1
10. 10
  Stahl, M.; Rarey, M. Detailed analysis of scoring functions for virtual screening J. Med. Chem. 2001, 44, 1035– 1042
  Google Scholar
  10
  Detailed Analysis of Scoring Functions for Virtual Screening
  Stahl, Martin; Rarey, Matthias
  Journal of Medicinal Chemistry (2001), 44 (7), 1035-1042CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  The authors present a comprehensive study of the performance of fast scoring functions for library docking using the program FlexX as the docking engine. Four scoring functions, among them two recently developed knowledge-based potentials, are evaluated on seven target proteins whose binding sites represent a wide range of size, form, and polarity. The results of these calcns. give valuable insight into strengths and weaknesses of current scoring functions. Furthermore, it is shown that a well-chosen combination of two of the tested scoring functions leads to a new, robust scoring scheme with superior performance in virtual screening.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXhsFKitro%253D&md5=9fe0ea156ae846868e710115e1d71500
11. 11
  Bissantz, C.; Folkers, G.; Rognan, D. Protein-based virtual screening of chemical databases. 1. Evaluation of different docking/scoring combinations J. Med. Chem. 2000, 43, 4759– 4767
  Google Scholar
  11
  Protein-Based Virtual Screening of Chemical Databases. 1. Evaluation of Different Docking/Scoring Combinations
  Bissantz, Caterina; Folkers, Gerd; Rognan, Didier
  Journal of Medicinal Chemistry (2000), 43 (25), 4759-4767CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Three different database docking programs (Dock, FlexX, Gold) have been used in combination with seven scoring functions (Chemscore, Dock, FlexX, Fresno, Gold, Pmf, Score) to assess the accuracy of virtual screening methods against two protein targets (thymidine kinase, estrogen receptor) of known three-dimensional structure. For both targets, it was generally possible to discriminate about 7 out of 10 true hits from a random database of 990 ligands. The use of consensus lists common to two or three scoring functions clearly enhances hit rates among the top 5% scorers from 10% (single scoring) to 25-40% (double scoring) and up to 65-70% (triple scoring). However, in all tested cases, no clear relationships could be found between docking and ranking accuracies. Moreover, predicting the abs. binding free energy of true hits was not possible whatever docking accuracy was achieved and scoring function used. As the best docking/consensus scoring combination varies with the selected target and the physicochem. of target-ligand interactions, we propose a two-step protocol for screening large databases: (i) screening of a reduced dataset contg. a few known ligands for deriving the optimal docking/consensus scoring scheme, (ii) applying the latter parameters to the screening of the entire database.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXotFymurY%253D&md5=ee74ac9a99e55759c2df27fd1db58010
12. 12
  Pham, T. A.; Jain, A. N. Parameter estimation for scoring protein–ligand interactions using negative training data J. Med. Chem. 2006, 49, 5856– 5868
  Google Scholar
  12
  Parameter Estimation for Scoring Protein-Ligand Interactions Using Negative Training Data
  Pham, Tuan A.; Jain, Ajay N.
  Journal of Medicinal Chemistry (2006), 49 (20), 5856-5868CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Surflex-Dock employs an empirically derived scoring function to rank putative protein-ligand interactions by flexible docking of small mols. to proteins of known structure. The scoring function employed by Surflex was developed purely on the basis of pos. data, comprising noncovalent protein-ligand complexes with known binding affinities. Consequently, scoring function terms for improper interactions received little wt. in parameter estn., and an ad hoc scheme for avoiding protein-ligand interpenetration was adopted. We present a generalized method for incorporating synthetically generated neg. training data, which allows for rigorous estn. of all scoring function parameters. Geometric docking accuracy remained excellent under the new parametrization. In addn., a test of screening utility covering a diverse set of 29 proteins and corresponding ligand sets showed improved performance. Maximal enrichment of true ligands over non-ligands exceeded 20-fold in over 80% of cases, with enrichment of greater than 100-fold in over 50% of cases.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXlsF2ls70%253D&md5=2296d0e658abfda21e3cc50d2a248e44
13. 13
  Kellenberger, E.; Rodrigo, J.; Muller, P.; Rognan, D. Comparative evaluation of eight docking tools for docking and virtual screening accuracy Proteins 2004, 57, 225– 242
  Google Scholar
  13
  Comparative evaluation of eight docking tools for docking and virtual screening accuracy
  Kellenberger, Esther; Rodrigo, Jordi; Muller, Pascal; Rognan, Didier
  Proteins: Structure, Function, and Bioinformatics (2004), 57 (2), 225-242CODEN: PSFBAF ISSN:. (Wiley-Liss, Inc.)
  Eight docking programs (DOCK, FLEXX, FRED, GLIDE, GOLD, SLIDE, SURFLEX, and QXP) that can be used for either single-ligand docking or database screening have been compared for their propensity to recover the x-ray pose of 100 small-mol.-wt. ligands, and for their capacity to discriminate known inhibitors of an enzyme (thymidine kinase) from randomly chosen "drug-like" mols. Interestingly, both properties are found to be correlated, since the tools showing the best docking accuracy (GLIDE, GOLD, and SURFLEX) are also the most successful in ranking known inhibitors in a virtual screening expt. Moreover, the current study pinpoints some physicochem. descriptors of either the ligand or its cognate protein-binding site that generally lead to docking/scoring inaccuracies.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXnvFaku7k%253D&md5=75a11a8f0603c819e4f4c8013b948686
14. 14
  Ferrara, P.; Gohlke, H.; Price, D. J.; Klebe, G.; Brooks, C. L., III. Assessing scoring functions for protein–ligand interactions J. Med. Chem. 2004, 47, 3032– 3047
  Google Scholar
  14
  Assessing Scoring Functions for Protein-Ligand Interactions
  Ferrara, Philippe; Gohlke, Holger; Price, Daniel J.; Klebe, Gerhard; Brooks, Charles L., III
  Journal of Medicinal Chemistry (2004), 47 (12), 3032-3047CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  An assessment of nine scoring functions commonly applied in docking using a set of 189 protein-ligand complexes is presented. The scoring functions include the CHARMm potential, the scoring function DrugScore, the scoring function used in AutoDock, the three scoring functions implemented in DOCK, as well as three scoring functions implemented in the CScore module in SYBYL (PMF, Gold, ChemScore). The authors evaluated the abilities of these scoring functions to recognize near-native configurations among a set of decoys and to rank binding affinities. Binding site decoys were generated by mol. dynamics with restraints. To investigate whether the scoring functions can also be applied for binding site detection, decoys on the protein surface were generated. The influence of the assignment of protonation states was probed by either assigning "std." protonation states to binding site residues or adjusting protonation states according to exptl. evidence. The role of solvation models in conjunction with CHARMm was explored in detail. These include a distance-dependent dielec. function, a generalized Born model, and the Poisson equation. The authors evaluated the effect of using a rigid receptor on the outcome of docking by generating all-pairs decoys ("cross-decoys") for six trypsin and seven HIV-1 protease complexes. The scoring functions perform well to discriminate near-native from misdocked conformations, with CHARMm, DOCK-energy, DrugScore, ChemScore, and AutoDock yielding recognition rates of around 80%. Significant degrdn. in performance is obsd. in going from decoy to cross-decoy recognition for CHARMm in the case of HIV-1 protease, whereas DrugScore and ChemScore, as well as CHARMm in the case of trypsin, show only small deterioration. In contrast, the prediction of binding affinities remains problematic for all of the scoring functions. ChemScore gives the highest correlation value with R2 = 0.51 for the set of 189 complexes and R2 = 0.43 for the set of 116 complexes that does not contain any of the complexes used to calibrate this scoring function. Neither a more accurate treatment of solvation nor a more sophisticated charge model for zinc improves the quality of the results. Improved modeling of the protonation states, however, leads to a better prediction of binding affinities in the case of the generalized Born and the Poisson continuum models used in conjunction with the CHARMm force field. The method can be used for drug discovery.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXjs1Sjsrk%253D&md5=a9f1e74194d3949aefadefb3266372d2
15. 15
  Huang, N.; Shoichet, B. K.; Irwin, J. J. Benchmarking sets for molecular docking J. Med. Chem. 2006, 49, 6789– 6801
  Google Scholar
  15
  Benchmarking Sets for Molecular Docking
  Huang, Niu; Shoichet, Brian K.; Irwin, John J.
  Journal of Medicinal Chemistry (2006), 49 (23), 6789-6801CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Ligand enrichment among top-ranking hits is a key metric of mol. docking. To avoid bias, decoys should resemble ligands phys., so that enrichment is not simply a sepn. of gross features, yet be chem. distinct from them, so that they are unlikely to be binders. We have assembled a directory of useful decoys (DUD), with 2950 ligands for 40 different targets. Every ligand has 36 decoy mols. that are phys. similar but topol. distinct, leading to a database of 98 266 compds. For most targets, enrichment was at least half a log better with uncorrected databases such as the MDDR than with DUD, evidence of bias in the former. These calcns. also allowed 40×40 cross-docking, where the enrichments of each ligand set could be compared for all 40 targets, enabling a specificity metric for the docking screens. DUD is freely available online as a benchmarking set for docking at http://blaster.docking.org/dud/.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28XhtFehurzI&md5=43af46968c2a6d66334922bf0caedbe1
16. 16
  Christofferson, A. J.; Huang, N. How to benchmark methods for structure-based virtual screening of large compound libraries. In Computational Drug Discovery and Design (Methods in Molecular Biology); 2011/12/21 ed.; Baron, R., Ed.; Springer Protocols: New York, 2012; Vol. 819, Chapter 13, pp 187– 195.
  Google Scholar
  There is no corresponding record for this reference.
17. 17
  Verdonk, M. L.; Berdini, V.; Hartshorn, M. J.; Mooij, W. T.; Murray, C. W.; Taylor, R. D.; Watson, P. Virtual screening using protein–ligand docking: avoiding artificial enrichment J. Chem. Inf. Comput. Sci. 2004, 44, 793– 806
  Google Scholar
  17
  Virtual screening using protein-ligand docking: Avoiding artificial enrichment
  Verdonk, Marcel L.; Berdini, Valerio; Hartshorn, Michael J.; Mooij, Wijnand T. M.; Murray, Christopher W.; Taylor, Richard D.; Watson, Paul
  Journal of Chemical Information and Computer Sciences (2004), 44 (3), 793-806CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
  This study addresses a no. of topical issues around the use of protein-ligand docking in virtual screening. We show that, for the validation of such methods, it is key to use focused libraries (contg. compds. with one-dimensional properties, similar to the actives), rather than "random" or "drug-like" libraries to test the actives against. We also show that, to obtain good enrichments, the docking program needs to produce reliable binding modes. We demonstrate how pharmacophores can be used to guide the dockings and improve enrichments, and we compare the performance of three consensus-ranking protocols against ranking based on individual scoring functions. Finally, we show that protein-ligand docking can be an effective aid in the screening for weak, fragment-like binders, which has rapidly become a popular strategy for hit identification. All results presented are based on carefully constructed virtual screening expts. against four targets, using the protein-ligand docking program GOLD.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXhtVGhs70%253D&md5=b0cd9e9a04ddebe401dee9744e2184cb
18. 18
  Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A. The maximal affinity of ligands Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 9997– 10002
  Google Scholar
  18
  The maximal affinity of ligands
  Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A.
  Proceedings of the National Academy of Sciences of the United States of America (1999), 96 (18), 9997-10002CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
  We explore the question of what are the best ligands for macromol. targets. A survey of exptl. data on a large no. of the strongest-binding ligands indicates that the free energy of binding increases with the no. of nonhydrogen atoms with an initial slope of ≈-1.5 kcal/mol (1 cal = 4.18 J) per atom. For ligands that contain more than 15 nonhydrogen atoms, the free energy of binding increases very little with relative mol. mass. This nonlinearity is largely ascribed to nonthermodynamic factors. An anal. of the dominant interactions suggests that van der Waals interactions and hydrophobic effects provide a reasonable basis for understanding binding affinities across the entire set of ligands. Interesting outliers that bind unusually strongly on a per atom basis include metal ions, covalently attached ligands, and a few well known complexes such as biotin-avidin.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK1MXlvFehu7s%253D&md5=46fe894b016f3041831a2f0b71f812b1
19. 19
  Fan, H.; Irwin, J. J.; Webb, B. M.; Klebe, G.; Shoichet, B. K.; Sali, A. Molecular Docking Screens Using Comparative Models of Proteins J. Chem. Inf. Model. 2009, 49, 2512– 2527
  Google Scholar
  19
  Molecular Docking Screens Using Comparative Models of Proteins
  Fan, Hao; Irwin, John J.; Webb, Benjamin M.; Klebe, Gerhard; Shoichet, Brian K.; Sali, Andrej
  Journal of Chemical Information and Modeling (2009), 49 (11), 2512-2527CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Two orders of magnitude more protein sequences can be modeled by comparative modeling than have been detd. by X-ray crystallog. and NMR spectroscopy. Investigators have nevertheless been cautious about using comparative models for ligand discovery because of concerns about model errors. We suggest how to exploit comparative models for mol. screens, based on docking against a wide range of crystallog. structures and comparative models with known ligands. To account for the variation in the ligand-binding pocket as it binds different ligands, we calc. "consensus" enrichment by ranking each library compd. by its best docking score against all available comparative models and/or modeling templates. For the majority of the targets, the consensus enrichment for multiple models was better than or comparable to that of the holo and apo X-ray structures. Even for single models, the models are significantly more enriching than the template structure if the template is paralogous and shares more than 25% sequence identity with the target.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXht1yntL7M&md5=11bd5284c6b43592906e45b6aea3019d
20. 20
  Repasky, M. P.; Murphy, R. B.; Banks, J. L.; Greenwood, J. R.; Tubert-Brohman, I.; Bhat, S.; Friesner, R. A. Docking performance of the glide program as evaluated on the Astex and DUD datasets: a complete set of glide SP results and selected results for a new scoring function integrating WaterMap and glide J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9575-9
  Google Scholar
  There is no corresponding record for this reference.
21. 21
  Brozell, S. R.; Mukherjee, S.; Balius, T. E.; Roe, D. R.; Case, D. A.; Rizzo, R. C. Evaluation of DOCK 6 as a pose generation and database enrichment tool J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9565-y
  Google Scholar
  There is no corresponding record for this reference.
22. 22
  Neves, M. A.; Totrov, M.; Abagyan, R. Docking and scoring with ICM: the benchmarking results and strategies for improvement J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9547-0
  Google Scholar
  There is no corresponding record for this reference.
23. 23
  Spitzer, R.; Jain, A. N. Surflex-Dock: docking benchmarks and real-world application J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-011-9533-y
  Google Scholar
  There is no corresponding record for this reference.
24. 24
  Schneider, N.; Hindle, S.; Lange, G.; Klein, R.; Albrecht, J.; Briem, H.; Beyer, K.; Claussen, H.; Gastreich, M.; Lemmen, C.; Rarey, M. Substantial improvements in large-scale redocking and screening using the novel HYDE scoring function J. Comput.-Aided Mol. Des. 2011, DOI: 10.1007/s10822-011-9531-0
  Google Scholar
  There is no corresponding record for this reference.
25. 25
  Liebeschuetz, J. W.; Cole, J. C.; Korb, O. Pose prediction and virtual screening performance of GOLD scoring functions in a standardized test J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9551-4
  Google Scholar
  There is no corresponding record for this reference.
26. 26
  Novikov, F. N.; Stroylov, V. S.; Zeifman, A. A.; Stroganov, O. V.; Kulkov, V.; Chilov, G. G. Lead Finder docking and virtual screening evaluation with Astex and DUD test sets J. Comput.-Aided Mol. Des. 2012, DOI: 10.1007/s10822-012-9549-y
  Google Scholar
  There is no corresponding record for this reference.
27. 27
  Good, A. C.; Oprea, T. I. Optimization of CAMD techniques 3. Virtual screening enrichment studies: a help or hindrance in tool selection? J. Comput.-Aided Mol. Des. 2008, 22, 169– 178
  Google Scholar
  27
  Optimization of CAMD techniques 3. Virtual screening enrichment studies: a help or hindrance in tool selection?
  Good, Andrew C.; Oprea, Tudor I.
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 169-178CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  Over the last few years many articles have been published in an attempt to provide performance benchmarks for virtual screening tools. While this research has imparted useful insights, the myriad variables controlling said studies place significant limits on results interpretability. Here we investigate the effects of these variables, including anal. of calcn. setup variation, the effect of target choice, active/decoy set selection (with particular emphasis on the effect of analog bias) and enrichment data interpretation. In addn. the optimization of the publicly available DUD benchmark sets through analog bias removal is discussed, as is their augmentation through the addn. of large diverse data sets collated using WOMBAT.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsLo%253D&md5=459cd3d09a88d5afc04a4a49e172fb76
28. 28
  Mackey, M. D.; Melville, J. L. Better than random? The chemotype enrichment problem J. Chem. Inf. Model. 2009, 49, 1154– 1162
  Google Scholar
  28
  Better than Random? The Chemotype Enrichment Problem
  Mackey, Mark D.; Melville, James L.
  Journal of Chemical Information and Modeling (2009), 49 (5), 1154-1162CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Chemotype enrichment is increasingly recognized as an important measure of virtual screening performance. However, little attention has been paid to producing metrics which can quantify chemotype retrieval. Here, we examine two different protocols for analyzing chemotype retrieval: "cluster averaging", where the contribution of each active to the scoring metric is proportional to the no. of other actives with the same chemotype, and "first found", where only the first active for a given chemotype contributes to the score. We demonstrate that this latter anal., common in the qual. anal. used in the current literature, has important drawbacks when combined with quant. metrics.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXltVCgs78%253D&md5=51b6efb624c3a383b7546997c11ab1e0
29. 29
  Hawkins, P. C.; Warren, G. L.; Skillman, A. G.; Nicholls, A. How to do an evaluation: pitfalls and traps J. Comput.-Aided Mol. Des. 2008, 22, 179– 190
  Google Scholar
  29
  How to do an evaluation: pitfalls and traps
  Hawkins, Paul C. D.; Warren, Gregory L.; Skillman, A. Geoffrey; Nicholls, Anthony
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 179-190CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  The recent literature is replete with papers evaluating computational tools (often those operating on 3D structures) for their performance in a certain set of tasks. Most commonly these papers compare a no. of docking tools for their performance in cognate re-docking (pose prediction) and/or virtual screening. Related papers have been published on ligand-based tools: pose prediction by conformer generators and virtual screening using a variety of ligand-based approaches. The reliability of these comparisons is critically affected by a no. of factors usually ignored by the authors, including bias in the datasets used in virtual screening, the metrics used to assess performance in virtual screening and pose prediction and errors in crystal structures used.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsLk%253D&md5=3e176941c4a526f3a97519ce02007792
30. 30
  Irwin, J. J. Community benchmarks for virtual screening J. Comput.-Aided Mol. Des. 2008, 22, 193– 199
  Google Scholar
  30
  Community benchmarks for virtual screening
  Irwin, John J.
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 193-199CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  Ligand enrichment among top-ranking hits is a key metric of virtual screening. To avoid bias, decoys should resemble ligands phys., so that enrichment is not attributable to simple differences of gross features. We therefore created a directory of useful decoys (DUD) by selecting decoys that resembled annotated ligands phys. but not topol. to benchmark docking performance. DUD has 2950 annotated ligands and 95,316 property-matched decoys for 40 targets. It is by far the largest and most comprehensive public data set for benchmarking virtual screening programs that I am aware of. This paper outlines several ways that DUD can be improved to provide better telemetry to investigators seeking to understand both the strengths and the weaknesses of current docking methods. I also highlight several pitfalls for the unwary: a risk of over-optimization, questions about chem. space, and the proper scope for using DUD. Careful attention to both the compn. of benchmarks and how they are used is essential to avoid being misled by overfitting and bias.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsb8%253D&md5=4972b768c915bc3e7ee05210dcf1b7a8
31. 31
  Mysinger, M. M.; Shoichet, B. K. Rapid context-dependent ligand desolvation in molecular docking J. Chem. Inf. Model. 2010, 50, 1561– 1573
  Google Scholar
  31
  Rapid Context-Dependent Ligand Desolvation in Molecular Docking
  Mysinger, Michael M.; Shoichet, Brian K.
  Journal of Chemical Information and Modeling (2010), 50 (9), 1561-1573CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  In structure-based screens for new ligands, a mol. docking algorithm must rapidly score many mols. in multiple configurations, accounting for both the ligand's interactions with receptor and its competing interactions with solvent. Here the authors explore a context-dependent ligand desolvation scoring term for mol. docking. The authors relate the Generalized-Born effective Born radii for every ligand atom to a fractional desolvation and then use this fraction to scale an atom-by-atom decompn. of the full transfer free energy. The fractional desolvation is precomputed on a scoring grid by numerically integrating over the vol. of receptor proximal to a ligand atom, weighted by distance. To test this method's performance, the authors dock ligands vs. property-matched decoys over 40 DUD targets. Context-dependent desolvation better enriches ligands compared to both the raw full transfer free energy penalty and compared to ignoring desolvation altogether, though the improvement is modest. More compellingly, the new method improves docking performance across receptor types. Thus, whereas entirely ignoring desolvation works best for charged sites and overpenalizing with full desolvation works well for neutral sites, the phys. more correct context-dependent ligand desolvation is competitive across both types of targets. The method also reliably discriminates ligands from highly charged mols., where ignoring desolvation performs poorly. Since this context-dependent ligand desolvation may be precalcd., it improves docking reliability with minimal cost to calcn. time and may be readily incorporated into any physics-based docking program.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXhtVGrsLvI&md5=5124d2c4618e07b4f1e41609c9e61362
32. 32
  Vogel, S. M.; Bauer, M. R.; Boeckler, F. M. DEKOIS: demanding evaluation kits for objective in silico screening—a versatile tool for benchmarking docking programs and scoring functions J. Chem. Inf. Model. 2011, 51, 2650– 2665
  Google Scholar
  32
  DEKOIS: Demanding Evaluation Kits for Objective in Silico Screening - A Versatile Tool for Benchmarking Docking Programs and Scoring Functions
  Vogel, Simon M.; Bauer, Matthias R.; Boeckler, Frank M.
  Journal of Chemical Information and Modeling (2011), 51 (10), 2650-2665CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  For widely applied in silico screening techniques success depends on the rational selection of an appropriate method. We herein present a fast, versatile, and robust method to construct demanding evaluation kits for objective in silico screening (DEKOIS). This automated process enables creating tailor-made decoy sets for any given sets of bioactives. It facilitates a target-dependent validation of docking algorithms and scoring functions helping to save time and resources. We have developed metrics for assessing and improving decoy set quality and employ them to investigate how decoy embedding affects docking. We demonstrate that screening performance is target-dependent and can be impaired by latent actives in the decoy set (LADS) or enhanced by poor decoy embedding. The presented method allows extending and complementing the collection of publicly available high quality decoy sets toward new target space. All present and future DEKOIS data sets will be made accessible at www.dekois.com.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtValsbnN&md5=e3f59bcd3f898dce1c116a325e914861
33. 33
  Wallach, I.; Lilien, R. Virtual decoy sets for molecular docking benchmarks J. Chem. Inf. Model. 2011, 51, 196– 202
  Google Scholar
  33
  Virtual Decoy Sets for Molecular Docking Benchmarks
  Wallach, Izhar; Lilien, Ryan
  Journal of Chemical Information and Modeling (2011), 51 (2), 196-202CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Virtual docking algorithms are often evaluated on their ability to sep. active ligands from decoy mols. The current state-of-the-art benchmark, the Directory of Useful Decoys (DUD), minimizes bias by including decoys from a library of synthetically feasible mols. that are phys. similar yet chem. dissimilar to the active ligands. We show that by ignoring synthetic feasibility, we can compile a benchmark that is comparable to the DUD and less biased with respect to phys. similarity.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXoslel&md5=3b58c7641cb2c665ce381081d8b1bf41
34. 34
  Gatica, E. A.; Cavasotto, C. N. Ligand and decoy sets for docking to G protein-coupled receptors J. Chem. Inf. Model. 2012, 52, 1– 6
  Google Scholar
  34
  Ligand and Decoy Sets for Docking to G Protein-Coupled Receptors
  Gatica, Edgar A.; Cavasotto, Claudio N.
  Journal of Chemical Information and Modeling (2012), 52 (1), 1-6CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  We compiled a G protein-coupled receptor (GPCR) ligand library (GLL) for 147 targets, selecting for each ligand 39 decoy mols., collected in the GPCR Decoy Database (GDD). Decoys were chosen ensuring a ligand-decoy similarity of six phys. properties, while enforcing ligand-decoy chem. dissimilarity. The performance in docking of the GDD was evaluated on 19 GPCRs, showing a marked decrease in enrichment compared to bias-uncorrected decoy sets. Both the GLL and GDD are freely available for the scientific community.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1ajsb3P&md5=a51e1b60afda728614a3bc1ceed79cfd
35. 35
  Cereto-Massague, A.; Guasch, L.; Valls, C.; Mulero, M.; Pujadas, G.; Garcia-Vallve, S. DecoyFinder: an easy-to-use python GUI application for building target-specific decoy sets Bioinformatics 2012, 28, 1661– 1662
  Google Scholar
  There is no corresponding record for this reference.
36. 36
  Rohrer, S. G.; Baumann, K. Maximum unbiased validation (MUV) data sets for virtual screening based on PubChem bioactivity data J. Chem. Inf. Model. 2009, 49, 169– 184
  Google Scholar
  36
  Maximum Unbiased Validation (MUV) Data Sets for Virtual Screening Based on PubChem Bioactivity Data
  Rohrer, Sebastian G.; Baumann, Knut
  Journal of Chemical Information and Modeling (2009), 49 (2), 169-184CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Refined nearest neighbor anal. was recently introduced for the anal. of virtual screening benchmark data sets. It constitutes a technique from the field of spatial statistics and provides a math. framework for the nonparametric anal. of mapped point patterns. Here, refined nearest neighbor anal. is used to design benchmark data sets for virtual screening based on PubChem bioactivity data. A workflow is devised that purges data sets of compds. active against pharmaceutically relevant targets from unselective hits. Topol. optimization using exptl. design strategies monitored by refined nearest neighbor anal. functions is applied to generate corresponding data sets of actives and decoys that are unbiased with regard to analog bias and artificial enrichment. These data sets provide a tool for Maximum Unbiased Validation (MUV) of virtual screening methods. The data sets and a software package implementing the MUV design workflow are freely available.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXptlOhtQ%253D%253D&md5=17805f49d995c4c082c3d07be7145a2d
37. 37
  Ripphausen, P.; Wassermann, A. M.; Bajorath, J. REPROVIS-DB: a benchmark system for ligand-based virtual screening derived from reproducible prospective applications J. Chem. Inf. Model. 2011, 51, 2467– 2473
  Google Scholar
  37
  REPROVIS-DB: A Benchmark System for Ligand-Based Virtual Screening Derived from Reproducible Prospective Applications
  Ripphausen, Peter; Wassermann, Anne Mai; Bajorath, Jurgen
  Journal of Chemical Information and Modeling (2011), 51 (10), 2467-2473CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Benchmark calcns. are essential for the evaluation of virtual screening (VS) methods. Typically, classes of known active compds. taken from the medicinal chem. literature are divided into ref. mols. (search templates) and potential hits that are added to background databases assumed to consist of compds. not sharing this activity. Then VS calcns. are carried out, and the recall of known active compds. is detd. However, conventional benchmarking is affected by a no. of problems that reduce its value for method evaluation. In addn. to often insufficient statistical validation and the lack of generally accepted evaluation stds., the artificial nature of typical benchmark settings is often criticized. Retrospective benchmark calcns. generally overestimate the potential of VS methods and do not scale with their performance in prospective applications. In order to provide addnl. opportunities for benchmarking that more closely resemble practical VS conditions, we have designed a publicly available compd. database (DB) of reproducible virtual screens (REPROVIS-DB) that organizes information from successful ligand-based VS applications including ref. compds., screening databases, compd. selection criteria, and exptl. confirmed hits. Using the currently available 25 hand-selected compd. data sets, one can attempt to reproduce successful virtual screens with other than the originally applied methods and assess their potential for practical applications.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXht1ait7fJ&md5=8439a3f6471f43afe159d08c28a5bc14
38. 38
  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– 1107
  Google Scholar
  38
  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
39. 39
  Bemis, G. W.; Murcko, M. A. The properties of known drugs. 1. Molecular frameworks J. Med. Chem. 1996, 39, 2887– 2893
  Google Scholar
  39
  The Properties of Known Drugs. 1. Molecular Frameworks
  Bemis, Guy W.; Murcko, Mark A.
  Journal of Medicinal Chemistry (1996), 39 (15), 2887-2893CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  To better understand the common features present in drug mols., we use shape description methods to analyze a database of com. available drugs and prep. a list of common drug shapes. A useful way of organizing this structural data is to group the atoms of each drug mol. into ring, linker, framework, and side chain atoms. On the basis of the two-dimensional mol. structures (without regard to atom type, hybridization, and bond order), there are 1179 different frameworks among the 5120 compds. analyzed. However, the shapes of half of the drugs in the database are described by the 32 most frequently occurring frameworks. This suggests that the diversity of shapes in the set of known drugs is extremely low. In our second method of anal., in which atom type, hybridization, and bond order are considered, more diversity is seen; there are 2506 different frameworks among the 5120 compds. in the database, and the most frequently occurring 42 frameworks account for only one-fourth of the drugs. We discuss the possible interpretations of these findings and the way they may be used to guide future drug discovery research.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK28XjvVejtro%253D&md5=5e2c4fdfea9434456a0cca83de4185b3
40. 40
  Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank Nucleic Acids Res. 2000, 28, 235– 242
  Google Scholar
  40
  The Protein Data Bank
  Berman, Helen M.; Westbrook, John; Feng, Zukang; Gilliland, Gary; Bhat, T. N.; Weissig, Helge; Shindyalov, Ilya N.; Bourne, Philip E.
  Nucleic Acids Research (2000), 28 (1), 235-242CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
  The Protein Data Bank (PDB; http://www.rcsb.org/pdb/)is the single worldwide archive of structural data of biol. macromols. This paper describes the goals of the PDB, the systems in place for data deposition and access, how to obtain further information, and near-term plans for the future development of the resource.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXhvVKjt7w%253D&md5=227fb393f754be2be375ab727bfd05dc
41. 41
  Apweiler, R.; Bairoch, A.; Wu, C. H.; Barker, W. C.; Boeckmann, B.; Ferro, S.; Gasteiger, E.; Huang, H.; Lopez, R.; Magrane, M.; Martin, M. J.; Natale, D. A.; O’Donovan, C.; Redaschi, N.; Yeh, L. S. UniProt: The Universal Protein Knowledgebase Nucleic Acids Res. 2004, 32, D115– D119
  Google Scholar
  41
  UniProt: the universal protein knowledgebase
  Apweiler, Rolf; Bairoch, Amos; Wu, Cathy H.; Barker, Winona C.; Boeckmann, Brigitte; Ferro, Serenella; Gasteiger, Elisabeth; Huang, Hongzhan; Lopez, Rodrigo; Magrane, Michele; Martin, Maria J.; Natale, Darren A.; O'Donovan, Claire; Redaschi, Nicole; Yeh, Lai-Su L.
  Nucleic Acids Research (2004), 32 (Database), D115-D119CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)
  To provide the scientific community with a single, centralized, authoritative resource for protein sequences and functional information, the Swiss-Prot, TrEMBL and PIR protein database activities have united to form the Universal Protein Knowledgebase (UniProt) consortium. Our mission is to provide a comprehensive, fully classified, richly and accurately annotated protein sequence knowledgebase, with extensive cross-refs. and query interfaces. The central database will have two sections, corresponding to the familiar Swiss-Prot (fully manually curated entries) and TrEMBL (enriched with automated classification, annotation and extensive cross-refs.). For convenient sequence searches, UniProt also provides several non-redundant sequence databases. The UniProt NREF (UniRef) databases provide representative subsets of the knowledgebase suitable for efficient searching. The comprehensive UniProt Archive (UniParc) is updated daily from many public source databases. The UniProt databases can be accessed online (http://www.uniprot.org) or downloaded in several formats (ftp://ftp.uniprot.org/pub). The scientific community is encouraged to submit data for inclusion in UniProt.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3sXhtVSru7vK&md5=4b692ffbb5c27bd9b4d2a8d0809e84e1
42. 42
  Irwin, J. J.; Shoichet, B. K.; Mysinger, M. M.; Huang, N.; Colizzi, F.; Wassam, P.; Cao, Y. Automated docking screens: a feasibility study J. Med. Chem. 2009, 52, 5712– 5720
  Google Scholar
  42
  Automated Docking Screens: A Feasibility Study
  Irwin, John J.; Shoichet, Brian K.; Mysinger, Michael M.; Huang, Niu; Colizzi, Francesco; Wassam, Pascal; Cao, Yiqun
  Journal of Medicinal Chemistry (2009), 52 (18), 5712-5720CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Mol. docking is the most practical approach to leverage protein structure for ligand discovery, but the technique retains important liabilities that make it challenging to deploy on a large scale. We have therefore created an expert system, DOCK Blaster, to investigate the feasibility of full automation. The method requires a PDB code, sometimes with a ligand structure, and from that alone can launch a full screen of large libraries. A crit. feature is self-assessment, which ests. the anticipated reliability of the automated screening results using pose fidelity and enrichment. Against common benchmarks, DOCK Blaster recapitulates the crystal ligand pose within 2 Å rmsd 50-60% of the time; inferior to an expert, but respectable. Half the time the ligand also ranked among the top 5% of 100 phys. matched decoys chosen on the fly. Further tests were undertaken culminating in a study of 7755 eligible PDB structures. In 1398 cases, the redocked ligand ranked in the top 5% of 100 property-matched decoys while also posing within 2 Å rmsd, suggesting that unsupervised prospective docking is viable. DOCK Blaster is available at http://blaster.docking.org.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXhtVOjurjN&md5=f302db1f77b87f26d1fa2fece55bca96
43. 43
  Powers, R. A.; Morandi, F.; Shoichet, B. K. Structure-based discovery of a novel, noncovalent inhibitor of AmpC beta-lactamase Structure 2002, 10, 1013– 1023
  Google Scholar
  43
  Structure-Based Discovery of a Novel, Noncovalent Inhibitor of AmpC β-Lactamase
  Powers, Rachel A.; Morandi, Federica; Shoichet, Brian K.
  Structure (Cambridge, MA, United States) (2002), 10 (7), 1013-1023CODEN: STRUE6; ISSN:0969-2126. (Cell Press)
  β-Lactamases are the most widespread resistance mechanisms to β-lactam antibiotics, and there is a pressing need for novel, non-β-lactam drugs. A database of over 200,000 compds. was docked to the active site of AmpC β-lactamase to identify potential inhibitors. Fifty-six compds. were tested, and three had Ki values of 650 μM or better. The best of these, 3-[(4-chloroanilino)sulfonyl]thiophene-2-carboxylic acid, was a competitive noncovalent inhibitor (Ki = 26 μM), which also reversed resistance to β-lactams in bacteria expressing AmpC. The structure of AmpC in complex with this compd. was detd. by x-ray crystallog. to 1.94 A and reveals that the inhibitor interacts with key active-site residues in sites targeted in the docking calcn. Indeed, the exptl. detd. conformation of the inhibitor closely resembles the prediction. The structure of the enzyme-inhibitor complex presents an opportunity to improve binding affinity in a novel series of inhibitors discovered by structure-based methods.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD38XlsVyjsrk%253D&md5=5f539484d08191f0c65626493fc1b262
44. 44
  Carlsson, J.; Yoo, L.; Gao, Z. G.; Irwin, J. J.; Shoichet, B. K.; Jacobson, K. A. Structure-based discovery of A2A adenosine receptor ligands J. Med. Chem. 2010, 53, 3748– 3755
  Google Scholar
  44
  Structure-Based Discovery of A2A Adenosine Receptor Ligands
  Carlsson, Jens; Yoo, Lena; Gao, Zhan-Guo; Irwin, John J.; Shoichet, Brian K.; Jacobson, Kenneth A.
  Journal of Medicinal Chemistry (2010), 53 (9), 3748-3755CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  The recent detn. of X-ray structures of pharmacol. relevant GPCRs has made these targets accessible to structure-based ligand discovery. Here we explore whether novel chemotypes may be discovered for the A2A adenosine receptor, based on complementarity to its recently detd. structure. The A2A adenosine receptor signals in the periphery and the CNS, with agonists explored as anti-inflammatory drugs and antagonists explored for neurodegenerative diseases. We used mol. docking to screen a 1.4 million compd. database against the X-ray structure computationally and tested 20 high-ranking, previously unknown mols. exptl. Of these 35% showed substantial activity with affinities between 200 nM and 9 μM. For the most potent of these new inhibitors, over 50-fold specificity was obsd. for the A2A vs. the related A1 and A3 subtypes. These high hit rates and affinities at least partly reflect the bias of com. libraries toward GPCR-like chemotypes, an issue that we attempt to investigate quant. Despite this bias, many of the most potent new ligands were novel, dissimilar from known ligands, providing new lead structures for modulation of this medically important target.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXkvFaqsL8%253D&md5=c36c941d52d2cec06387d79c4c423d46
45. 45
  Carlsson, J.; Coleman, R. G.; Setola, V.; Irwin, J. J.; Fan, H.; Schlessinger, A.; Sali, A.; Roth, B. L.; Shoichet, B. K. Ligand discovery from a dopamine D3 receptor homology model and crystal structure Natre Chem. Biol. 2011, 7, 769– 778
  Google Scholar
  45
  Ligand discovery from a dopamine D3 receptor homology model and crystal structure
  Carlsson, Jens; Coleman, Ryan G.; Setola, Vincent; Irwin, John J.; Fan, Hao; Schlessinger, Avner; Sali, Andrej; Roth, Bryan L.; Shoichet, Brian K.
  Nature Chemical Biology (2011), 7 (11), 769-778CODEN: NCBABT; ISSN:1552-4450. (Nature Publishing Group)
  G protein-coupled receptors (GPCRs) are intensely studied as drug targets and for their role in signaling. With the detn. of the first crystal structures, interest in structure-based ligand discovery increased. Unfortunately, for most GPCRs no exptl. structures are available. The detn. of the D3 receptor structure and the challenge to the community to predict it enabled a fully prospective comparison of ligand discovery from a modeled structure vs. that of the subsequently released crystal structure. Over 3.3 million mols. were docked against a homol. model, and 26 of the highest ranking were tested for binding. Six had affinities ranging from 0.2 to 3.1 μM. Subsequently, the crystal structure was released and the docking screen repeated. Of the 25 compds. selected, five had affinities ranging from 0.3 to 3.0 μM. One of the new ligands from the homol. model screen was optimized for affinity to 81 nM. The feasibility of docking screens against modeled GPCRs more generally is considered.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtFylsLvP&md5=7242aa512079a348678b309d352ecdb9
46. 46
  Irwin, J. J.; Shoichet, B. K. ZINC—a free database of commercially available compounds for virtual screening J. Chem. Inf. Model. 2005, 45, 177– 182
  Google Scholar
  46
  ZINC - A Free Database of Commercially Available Compounds for Virtual Screening
  Irwin, John J.; Shoichet, Brian K.
  Journal of Chemical Information and Computer Sciences (2005), 45 (1), 177-182CODEN: JCISD8; ISSN:0095-2338. (American Chemical Society)
  A crit. barrier to entry into structure-based virtual screening is the lack of a suitable, easy to access database of purchasable compds. We have therefore prepd. a library of 727 842 mols., each with 3D structure, using catalogs of compds. from vendors (the size of this library continues to grow). The mols. have been assigned biol. relevant protonation states and are annotated with properties such as mol. wt., calcd. LogP, and no. of rotatable bonds. Each mol. in the library contains vendor and purchasing information and is ready for docking using a no. of popular docking programs. Within certain limits, the mols. are prepd. in multiple protonation states and multiple tautomeric forms. In one format, multiple conformations are available for the mols. This database is available for free download (http://zinc.docking.org) in several common file formats including SMILES, mol2, 3D SDF, and DOCK flexibase format. A Web-based query tool incorporating a mol. drawing interface enables the database to be searched and browsed and subsets to be created. Users can process their own mols. by uploading them to a server. Our hope is that this database will bring virtual screening libraries to a wide community of structural biologists and medicinal chemists.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXhtVOjt77J&md5=e3892b7dc8608b17a3e63541a5ed60e6
47. 47
  Velankar, S.; McNeil, P.; Mittard-Runte, V.; Suarez, A.; Barrell, D.; Apweiler, R.; Henrick, K. E-MSD: an integrated data resource for bioinformatics Nucleic Acids Res. 2005, 33, D262– 265
  Google Scholar
  There is no corresponding record for this reference.
48. 48
  Hawkins, P. C.; Skillman, A. G.; Nicholls, A. Comparison of shape-matching and docking as virtual screening tools J. Med. Chem. 2007, 50, 74– 82
  Google Scholar
  48
  Comparison of Shape-Matching and Docking as Virtual Screening Tools
  Hawkins, Paul C. D.; Skillman, A. Geoffrey; Nicholls, Anthony
  Journal of Medicinal Chemistry (2007), 50 (1), 74-82CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Ligand docking is a widely used approach in virtual screening. In recent years a large no. of publications have appeared in which docking tools are compared and evaluated for their effectiveness in virtual screening against a wide variety of protein targets. These studies have shown that the effectiveness of docking in virtual screening is highly variable due to a large no. of possible confounding factors. Another class of method that has shown promise in virtual screening is the shape-based, ligand-centric approach. Several direct comparisons of docking with the shape-based tool ROCS have been conducted using data sets from some of these recent docking publications. The results show that a shape-based, ligand-centric approach is more consistent than, and often superior to, the protein-centric approach taken by docking.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28Xhtlansb%252FF&md5=6f97f5c0cc092b4e225f7c2656c1bcf6
49. 49
  Teotico, D. G.; Babaoglu, K.; Rocklin, G. J.; Ferreira, R. S.; Giannetti, A. M.; Shoichet, B. K. Docking for fragment inhibitors of AmpC beta-lactamase Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 7455– 7460
  Google Scholar
  49
  Docking for fragment inhibitors of AmpC 83-lactamase
  Teotico, Denise G.; Babaoglu, Kerim; Rocklin, Gabriel J.; Ferreira, Rafaela S.; Giannetti, Anthony M.; Shoichet, Brian K.
  Proceedings of the National Academy of Sciences of the United States of America (2009), 106 (18), 7455-7460CODEN: PNASA6; ISSN:0027-8424. (National Academy of Sciences)
  Fragment screens for new ligands have had wide success, not withstanding their constraint to libraries of 1000-10,000 mols. Larger libraries would be addressable were mol. docking reliable for fragment screens, but this has not been widely accepted. To investigate docking's ability to prioritize fragments, a library of >137,000 such mols. were docked against the structure of β-lactamase. Forty-eight fragments highly ranked by docking were acquired and tested; 23 had Ki values ranging from 0.7 to 9.2 mM. X-ray crystal structures of the enzyme-bound complexes were detd. for 8 of the fragments. For 4, the correspondence between the predicted and exptl. structures was high (RMSD between 1.2 and 1.4 Å), whereas for another 2, the fidelity was lower but retained most key interactions (RMSD 2.4-2.6 Å). Two of the 8 fragments adopted very different poses in the active site owing to enzyme conformational changes. The 48% hit rate of the fragment docking compares very favorably with "lead-like" docking and high-throughput screening against the same enzyme. To understand this, we investigated the occurrence of the fragment scaffolds among larger, lead-like mols. Approx. 1% of com. available fragments contain these inhibitors whereas only 10-7% of lead-like mols. do. This suggests that many more chemotypes and combinations of chemotypes are present among fragments than are available among lead-like mols., contributing to the higher hit rates. The ability of docking to prioritize these fragments suggests that the technique can be used to exploit the better chemotype coverage that exists at the fragment level.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXmt1Khtrg%253D&md5=d2df01432d5381c226e8aa7e008457b5
50. 50
  Tondi, D.; Morandi, F.; Bonnet, R.; Costi, M. P.; Shoichet, B. K. Structure-based optimization of a non-beta-lactam lead results in inhibitors that do not up-regulate beta-lactamase expression in cell culture J. Am. Chem. Soc. 2005, 127, 4632– 4639
  Google Scholar
  50
  Structure-Based Optimization of a Non-β-lactam Lead Results in Inhibitors That Do Not Up-Regulate β-Lactamase Expression in Cell Culture
  Tondi, Donatella; Morandi, Federica; Bonnet, Richard; Costi, M. Paola; Shoichet, Brian K.
  Journal of the American Chemical Society (2005), 127 (13), 4632-4639CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
  Bacterial expression of β-lactamases is the most widespread resistance mechanism to β-lactam antibiotics, such as penicillins and cephalosporins. There is a pressing need for novel, non-β-lactam inhibitors of these enzymes. One previously discovered novel inhibitor of the β-lactamase AmpC (I) has several favorable properties: it is chem. dissimilar to β-lactams and is a noncovalent, competitive inhibitor of the enzyme. However, at 26 μM its activity is modest. Using the x-ray structure of the AmpC/I complex as a template, 14 analogs were designed and synthesized. The most active of these, (II), had a Ki of 1 μM, 26-fold better than the lead. To understand the origins of this improved activity, the structures of AmpC in complex with compd. II and an analog were detd. by x-ray crystallog. to 1.97 and 1.96 Å, resp. II was active in cell culture, reversing resistance to the third generation cephalosporin ceftazidime in bacterial pathogens expressing AmpC. In contrast to β-lactam-based inhibitors clavulanate and cefoxitin, compd. II did not up-regulate β-lactamase expression in cell culture but simply inhibited the enzyme expressed by the resistant bacteria. Its escape from this resistance mechanism derives from its dissimilarity to β-lactam antibiotics.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXit1Krsbk%253D&md5=ddad35ac3e6fb96cf65e12cd807f271c
51. 51
  Graves, A. P.; Brenk, R.; Shoichet, B. K. Decoys for docking J. Med. Chem. 2005, 48, 3714– 3728
  Google Scholar
  51
  Decoys for Docking
  Graves, Alan P.; Brenk, Ruth; Shoichet, Brian K.
  Journal of Medicinal Chemistry (2005), 48 (11), 3714-3728CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)
  Mol. docking is widely used to predict novel lead compds. for drug discovery. Success depends on the quality of the docking scoring function, among other factors. An imperfect scoring function can mislead by predicting incorrect ligand geometries or by selecting nonbinding mols. over true ligands. These false-pos. hits may be considered "decoys". Although these decoys are frustrating, they potentially provide important tests for a docking algorithm; the more subtle the decoy, the more rigorous the test. Indeed, decoy databases have been used to improve protein structure prediction algorithms and protein-protein docking algorithms. Here, we describe 20 geometric decoys in five enzymes and 166 "hit list" decoys-i.e., mols. predicted to bind by our docking program that were tested and found not to do so - for β-lactamase and two cavity sites in lysozyme. Esp. in the cavity sites, which are very simple, these decoys highlight particular weaknesses in our scoring function. We also consider the performance of five other widely used docking scoring functions against our geometric and hit list decoys. Intriguingly, whereas many of these other scoring functions performed better on the geometric decoys, they typically performed worse on the hit list decoys, often highly ranking mols. that seemed to poorly complement the model sites. Several of these "hits" from the other scoring functions were tested exptl. and found, in fact, to be decoys. Collectively, these decoys provide a tool for the development and improvement of mol. docking scoring functions. Such improvements may, in turn, be rapidly tested exptl. against these and related exptl. systems, which are well-behaved in assays and for structure detn.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXjvFKjsbs%253D&md5=1797b501c1777dc94553701d45c822f0
52. 52
  Hawkins, P. C.; Skillman, A. G.; Warren, G. L.; Ellingson, B. A.; Stahl, M. T. Conformer generation with OMEGA: algorithm and validation using high quality structures from the Protein Data Bank and Cambridge Structural Database J. Chem. Inf. Model. 2010, 50, 572– 584
  Google Scholar
  52
  Conformer Generation with OMEGA: Algorithm and Validation Using High Quality Structures from the Protein Databank and Cambridge Structural Database
  Hawkins, Paul C. D.; Skillman, A. Geoffrey; Warren, Gregory L.; Ellingson, Benjamin A.; Stahl, Matthew T.
  Journal of Chemical Information and Modeling (2010), 50 (4), 572-584CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)
  Here, we present the algorithm and validation for OMEGA, a systematic, knowledge-based conformer generator. The algorithm consists of three phases: assembly of an initial 3D structure from a library of fragments; exhaustive enumeration of all rotatable torsions using values drawn from a knowledge-based list of angles, thereby generating a large set of conformations; and sampling of this set by geometric and energy criteria. Validation of conformer generators like OMEGA has often been undertaken by comparing computed conformer sets to exptl. mol. conformations from crystallog., usually from the Protein Databank (PDB). Such an approach is fraught with difficulty due to the systematic problems with small mol. structures in the PDB. Methods are presented to identify a diverse set of small mol. structures from cocomplexes in the PDB that has maximal reliability. A challenging set of 197 high quality, carefully selected ligand structures from well-solved models was obtained using these methods. This set will provide a sound basis for comparison and validation of conformer generators in the future. Validation results from this set are compared to the results using structures of a set of druglike mols. extd. from the Cambridge Structural Database (CSD). OMEGA is found to perform very well in reproducing the crystallog. conformations from both these data sets using two complementary metrics of success.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3cXjtlaisrY%253D&md5=fb87ecc9c51eddef63b41fffcd9babee
53. 53
  Jain, A. N. Bias, reporting, and sharing: computational evaluations of docking methods J. Comput.-Aided Mol. Des. 2008, 22, 201– 212
  Google Scholar
  53
  Bias, reporting, and sharing: computational evaluations of docking methods
  Jain, Ajay N.
  Journal of Computer-Aided Molecular Design (2008), 22 (3-4), 201-212CODEN: JCADEQ; ISSN:0920-654X. (Springer)
  Computational methods for docking ligands to protein binding sites have become ubiquitous in drug discovery. Despite the age of the field, no stds. have been established with respect to methodol. evaluation of docking accuracy, virtual screening utility, or scoring accuracy. There are crit. issues relating to data sharing, data set design and prepn., and statistical reporting that have an impact on the degree to which a report will translate into real-world performance. These issues also have an impact on whether there is a transparent relationship between methodol. changes and reported performance improvements. This paper presents detailed examples of pitfalls in each area and makes recommendations as to best practices.
  >> More from SciFinder ^®
  https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXjsFOnsbg%253D&md5=5ec23dce64722830afa856002acd6cf6
Supporting Information
Supporting Information
ARTICLE SECTIONS
Jump To
Figure showing DUD-E workflows, while tables provide detailed target-by-target data and tab delimited text files provide the raw data. This material is available free of charge via the Internet at http://pubs.acs.org.
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 [ 5MB]

All Types

Directory of Useful Decoys, Enhanced (DUD-E): Better Ligands and Decoys for Better Benchmarking

Article Views

Altmetric

Citations

Abstract

Introduction

Results

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Mineralocorticoid Receptor (MCR)

Figure 6

Thyroid Hormone Receptor β1 (THB)

Serine/Threonine-Protein Kinase AKT (AKT1)

Discussion

Balancing Ligands and Decoys for Enrichment

Online Tools for Automated Generation of Further Ligand and Decoy Sets

Applications to Docking Optimization and Testing

Methods

ChEMBL and RCSB PDB Data Extraction

Target Selection Docking

Target Preparation

Ligand Preparation

Ligand Clustering

Automated Decoy Generation

Original DUD Comparison

Docking Calculations

Docking Metrics

Supporting Information

Terms & Conditions

Author Information

Acknowledgment

Abbreviations Used

References

Cited By

Abstract

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

References

Supporting Information

Supporting Information

Terms & Conditions

STEP 1:

STEP 2: