Predicting Total Drug Clearance and Volumes of Distribution Using the Machine Learning-Mediated Multimodal Method through the Imputation of Various Nonclinical DataClick to copy article linkArticle link copied!
- Hiroaki Iwata*Hiroaki Iwata*Email: [email protected]Graduate School of Medicine, Kyoto University, 53 Shogoin-kawaharacho, Sakyo-ku, Kyoto 606-8507, JapanMore by Hiroaki Iwata
- Tatsuru MatsuoTatsuru MatsuoFujitsu Ltd., 4-1-1 Kamikodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa 211-8588, JapanMore by Tatsuru Matsuo
- Hideaki MamadaHideaki MamadaDMPK Research Laboratories, Central Pharmaceutical Research Institute, Japan Tobacco Inc., 1-1, Murasaki-cho, Takatsuki, Osaka 569-1125, JapanMore by Hideaki Mamada
- Takahisa MotomuraTakahisa MotomuraCentral Pharmaceutical Research Institute, Japan Tobacco Inc., 1-1, Murasaki-cho, Takatsuki, Osaka 569-1125, JapanMore by Takahisa Motomura
- Mayumi MatsushitaMayumi MatsushitaFujitsu Ltd., 4-1-1 Kamikodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa 211-8588, JapanMore by Mayumi Matsushita
- Takeshi FujiwaraTakeshi FujiwaraGraduate School of Medicine, Kyoto University, 53 Shogoin-kawaharacho, Sakyo-ku, Kyoto 606-8507, JapanMore by Takeshi Fujiwara
- Kazuya MaedaKazuya MaedaGraduate School of Pharmaceutical Sciences, Department of Molecular Pharmacokinetics, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, JapanMore by Kazuya Maeda
- Koichi Handa*Koichi Handa*Email: [email protected]Toxicology & DMPK Research Department, Teijin Institute for Bio-medical Research, Teijin Pharma Limited, 4-3-2 Asahigaoka, Hino-shi, Tokyo 191-8512, JapanMore by Koichi Handa
Abstract
Pharmacokinetic research plays an important role in the development of new drugs. Accurate predictions of human pharmacokinetic parameters are essential for the success of clinical trials. Clearance (CL) and volume of distribution (Vd) are important factors for evaluating pharmacokinetic properties, and many previous studies have attempted to use computational methods to extrapolate these values from nonclinical laboratory animal models to human subjects. However, it is difficult to obtain sufficient, comprehensive experimental data from these animal models, and many studies are missing critical values. This means that studies using nonclinical data as explanatory variables can only apply a small number of compounds to their model training. In this study, we perform missing-value imputation and feature selection on nonclinical data to increase the number of training compounds and nonclinical datasets available for these kinds of studies. We could obtain novel models for total body clearance (CLtot) and steady-state Vd (Vdss) (CLtot: geometric mean fold error [GMFE], 1.92; percentage within 2-fold error, 66.5%; Vdss: GMFE, 1.64; percentage within 2-fold error, 71.1%). These accuracies were comparable to the conventional animal scale-up models. Then, this method differs from animal scale-up methods because it does not require animal experiments, which continue to become more strictly regulated as time passes.
This publication is licensed under
License Summary*
You are free to share(copy and redistribute) this article in any medium or format within the parameters below:
Creative Commons (CC): This is a Creative Commons license.
Attribution (BY): Credit must be given to the creator.
Non-Commercial (NC): Only non-commercial uses of the work are permitted.
No Derivatives (ND): Derivative works may be created for non-commercial purposes, but sharing is prohibited.
*Disclaimer
This summary highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. Carefully review the actual license before using these materials.
License Summary*
You are free to share(copy and redistribute) this article in any medium or format within the parameters below:
Creative Commons (CC): This is a Creative Commons license.
Attribution (BY): Credit must be given to the creator.
Non-Commercial (NC): Only non-commercial uses of the work are permitted.
No Derivatives (ND): Derivative works may be created for non-commercial purposes, but sharing is prohibited.
*Disclaimer
This summary highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. Carefully review the actual license before using these materials.
License Summary*
You are free to share(copy and redistribute) this article in any medium or format within the parameters below:
Creative Commons (CC): This is a Creative Commons license.
Attribution (BY): Credit must be given to the creator.
Non-Commercial (NC): Only non-commercial uses of the work are permitted.
No Derivatives (ND): Derivative works may be created for non-commercial purposes, but sharing is prohibited.
*Disclaimer
This summary highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. Carefully review the actual license before using these materials.
License Summary*
You are free to share(copy and redistribute) this article in any medium or format within the parameters below:
Creative Commons (CC): This is a Creative Commons license.
Attribution (BY): Credit must be given to the creator.
Non-Commercial (NC): Only non-commercial uses of the work are permitted.
No Derivatives (ND): Derivative works may be created for non-commercial purposes, but sharing is prohibited.
*Disclaimer
This summary highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. Carefully review the actual license before using these materials.
License Summary*
You are free to share(copy and redistribute) this article in any medium or format within the parameters below:
Creative Commons (CC): This is a Creative Commons license.
Attribution (BY): Credit must be given to the creator.
Non-Commercial (NC): Only non-commercial uses of the work are permitted.
No Derivatives (ND): Derivative works may be created for non-commercial purposes, but sharing is prohibited.
*Disclaimer
This summary highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. Carefully review the actual license before using these materials.
Introduction
Materials and Methods
Workflow
Figure 1
Figure 1. Workflow of our novel human CLtot and Vdss prediction method. (A) CLtot analysis flow. (i) There were 741 compounds with human CLtot data and 46 that had values for all 11 features. (ii) All feature values were estimated via prediction using ADMEWORKS. (iii) Feature extraction was performed using XGBoost or Random Forest, and a prediction model was constructed. (B) Vdss analysis flow. (i) There were 751 compounds with human Vdss data and 46 that had values for all 11 features. (ii) All feature values were estimated via prediction using ADMEWORKS. (iii) Feature extraction was performed using XGBoost or Random Forest, and a prediction model was constructed.
Gathering Chemical Compounds, Nonclinical Data, and Data Preprocessing
feature | number of compounds | source |
---|---|---|
human CLtot | 741 | JCP2013, ChEMBL23 |
rat CLtot | 387 | JCP2013, ChEMBL23 |
dog CLtot | 284 | JCP2013, ChEMBL23 |
monkey CLtot | 129 | JCP2013, ChEMBL23 |
human Vdss | 751 | JCP2013, ChEMBL23 |
rat Vdss | 351 | JCP2013, ChEMBL23 |
dog Vdss | 274 | JCP2013, ChEMBL23 |
monkey Vdss | 125 | JCP2013, ChEMBL23 |
human fu | 577 | JCP2013, ChEMBL23 |
rat fu | 237 | JCP2013, ChEMBL23 |
dog fu | 179 | JCP2013, ChEMBL23 |
monkey fu | 88 | JCP2013, ChEMBL23 |
pKa acid | 334 | Pubchem, DrugBank |
pKa base | 335 | Pubchem, DrugBank |
solubility | 339 | Pubchem, DrugBank |
caco-2 permeability | 307 | Pubchem, DrugBank |
Missing-Value Imputation Using ADMEWORKS
Feature Selection and Prediction Model Construction
Feature Selection
Deep Tensor Model
Multimodal Deep Tensor Model
Figure 2
Figure 2. Overview of the multimodal Deep Tensor model.
XGBoost Model
Animal Scale-Up and Conventional Machine Learning Methods
Performance Evaluation

Results
Evaluation of the Usefulness of Missing-Value Imputation
method | GMFE | % of 2-fold error | |
---|---|---|---|
CLtot | 695 compounds | 2.53 | 45.7 |
XGBoost: only CS | |||
343 compounds | 2.15 | 52.2 | |
XGBoost: CS + rat CLtot | |||
695 compounds | 2.09 | 54.3 | |
XGBoost: CS + rat CLtot imputed | |||
695 compounds | 2.44 | 45.7 | |
Deep Tensor: only CS | |||
343 compounds | 2.15 | 54.8 | |
Deep Tensor: CS + rat CLtot | |||
695 compounds | 2.09 | 54.3 | |
Deep Tensor: CS + rat CLtot imputed | |||
Vdss | 706 compounds | 1.66 | 82.2 |
XGBoost: only CS | |||
306 compounds | 1.72 | 75.6 | |
XGBoost: CS + rat Vdss | |||
706 compounds | 1.73 | 68.9 | |
XGBoost: CS + rat Vdss imputed | |||
706 compounds | 1.85 | 62.2 | |
Deep Tensor: only CS | |||
306 compounds | 1.89 | 56.9 | |
Deep Tensor: CS + rat Vdss | |||
706 compounds | 1.75 | 64.4 | |
Deep Tensor: CS + rat Vdss imputed |
Improving Accuracy Using Feature Selection
methoda | GMFEb | % of 2-fold error | |
---|---|---|---|
CLtot | SSS rat | 2.36 | 43.5 |
SSS dog | 2.30 | 39.1 | |
SSS monkey | 1.93 | 58.7 | |
SA | 2.33 | 45.7 | |
FCIM | 1.99 | 52.2 | |
XGBoost: only CS | 2.40 | 50.0 | |
XGBoost: CS + 11 features | 2.06 | 58.7 | |
XGBoost: CS + selected features | 1.98 | 50.0 | |
Deep Tensor: only CS | 2.44 | 45.7 | |
Deep Tensor: CS + 11 features | 2.11 | 52.2 | |
Deep Tensor: CS + selected features | 1.92 | 66.5 | |
Vdss | SSS rat | 1.91 | 62.2 |
SSS dog | 1.93 | 71.1 | |
SSS monkey | 1.60 | 80.0 | |
SA | 2.07 | 68.9 | |
Øie–Tozer | 1.46 | 84.4 | |
XGBoost: only CS | 1.70 | 77.8 | |
XGBoost: CS + 11 features | 1.64 | 71.1 | |
XGBoost: CS + selected features | 1.66 | 71.1 | |
Deep Tensor: only CS | 1.85 | 62.2 | |
Deep Tensor: CS + 11 features | 1.75 | 69.8 | |
Deep Tensor: CS + selected features | 1.74 | 74.2 |
SSS: single-species allometric scaling; SA: simple allometry; FCIM: fu-corrected intercept method; CS: chemical structure.
GMFE: geometric mean fold error.
Discussion
Missing-Value Imputation (NA Imputation)
Selected Explanatory Variables
Comparison with Animal Scale-Up Methods
Conclusions
Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.2c00318.
The data used in this paper are listed in Supporting_information_dataset.xlsx (PDF)
Terms & Conditions
Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.
Acknowledgments
This research was conducted as part of the activities of the Life Intelligence Consortium (LINC). The authors thank Dr. Yasushi Okuno from the Graduate School of Medicine at Kyoto University for supporting their research activities at the LINC.
CL | clearance |
CLtot | total clearance |
Vd | volume of distribution |
Vdss | steady-state volume of distribution |
SSS | single-species allometric scaling |
SA | simple allometry |
FCIM | fraction-unbound corrected intercept method |
CS | chemical structure |
GMFE | geometric mean fold error |
CYP | cytochrome P450 |
References
This article references 36 other publications.
- 1Ballard, P.; Brassil, P.; Bui, K. H.; Dolgos, H.; Petersson, C.; Tunek, A.; Webborn, P. J. The right compound in the right assay at the right time: an integrated discovery DMPK strategy. Drug Metab. Rev. 2012, 44, 224– 252, DOI: 10.3109/03602532.2012.691099Google Scholar1https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XhtVelsLzF&md5=105fbbcfba6213caaadf41c8b10cf408The right compound in the right assay at the right time: an integrated discovery DMPK strategyBallard, Peter; Brassil, Patrick; Bui, Khanh H.; Dolgos, Hugues; Petersson, Carl; Tunek, Anders; Webborn, Peter J. H.Drug Metabolism Reviews (2012), 44 (3), 224-252CODEN: DMTRAR; ISSN:0360-2532. (Informa Healthcare)A review. The high rate of attrition during drug development and its assocd. high research and development (R&D) cost have put pressure on pharmaceutical companies to ensure that candidate drugs going to clin. testing have the appropriate quality such that the biol. hypothesis could be evaluated. To help achieve this ambition, drug metab. and pharmacokinetic (DMPK) science and increasing investment have been deployed earlier in the R&D process. To gain max. return on investment, it is essential that DMPK concepts are both appropriately integrated into the compd. design process and that compd. selection is focused on accurate prediction of likely outcomes in patients. This article describes key principles that underpin the contribution of DMPK science for small-mol. research based on 15 years of discovery support in a major pharmaceutical company. It does not aim to describe the breadth and depth of DMPK science, but more the practical application for decision making in real-world situations.
- 2Andrade, E. L.; Bento, A. F.; Cavalli, J.; Oliveira, S. K.; Schwanke, R. C.; Siqueira, J. M.; Freitas, C. S.; Marcon, R.; Calixto, J. B. Non-clinical studies in the process of new drug development - Part II: Good laboratory practice, metabolism, pharmacokinetics, safety and dose translation to clinical studies. Braz. J. Med. Biol. Res. 2016, 49, e5646 DOI: 10.1590/1414-431x20165646Google Scholar2https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXhsFarsbg%253D&md5=c76a3d9eb5b8844a77f26a705062578fNon-clinical studies in the process of new drug development - part II: good laboratory practice, metabolism, pharmacokinetics, safety and dose translation to clinical studiesAndrade, E. L.; Bento, A. F.; Cavalli, J.; Oliveir, S. K.; Schwank, R. C.; Siqueir, J. M.; Freita, C. S.; Marcon, R.; Calixto, J. B.Brazilian Journal of Medical and Biological Research (2016), 49 (12), e5646/1-e5646/19CODEN: BJMRDK; ISSN:0100-879X. (Associacao Brasileira de Divulgacao Cientifica)The process of drug development involves non-clin. and clin. studies. Non-clin. studies are conducted using different protocols including animal studies, which mostly follow the Good Lab. Practice (GLP) regulations. During the early preclin. development process, also known as Go/No-Go decision, a drug candidate needs to pass through several steps, such as detn. of drug availability (studies on pharmacokinetics), absorption, distribution, metab. and elimination (ADME) and preliminary studies that aim to investigate the candidate safety including genotoxicity, mutagenicity, safety pharmacol. and general toxicol. These preliminary studies generally do not need to comply with GLP regulations. These studies aim at investigating the drug safety to obtain the first information about its tolerability in different systems that are relevant for further decisions. There are, however, other studies that should be performed according to GLP stds. and are mandatory for the safe exposure to humans, such as repeated dose toxicity, genotoxicity and safety pharmacol. These studies must be conducted before the Investigational New Drug (IND) application. The package of non-clin. studies should cover all information needed for the safe transposition of drugs from animals to humans, generally based on the non-obsd. adverse effect level (NOAEL) obtained from general toxicity studies. After IND approval, other GLP expts. for the evaluation of chronic toxicity, reproductive and developmental toxicity, carcinogenicity and genotoxicity, are carried out during the clin. phase of development. However, the necessity of performing such studies depends on the new drug clin. application purpose.
- 3Lombardo, F.; Jing, Y. In Silico Prediction of Volume of Distribution in Humans. Extensive Data Set and the Exploration of Linear and Nonlinear Methods Coupled with Molecular Interaction Fields Descriptors. J. Chem. Inf. Model. 2016, 56, 2042– 2052, DOI: 10.1021/acs.jcim.6b00044Google Scholar3https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XhsVOrtLzI&md5=b99a25fcd384ecc777f9190f2061e49aIn Silico Prediction of Volume of Distribution in Humans. Extensive Data Set and the Exploration of Linear and Nonlinear Methods Coupled with Molecular Interaction Fields DescriptorsLombardo, Franco; Jing, YankangJournal of Chemical Information and Modeling (2016), 56 (10), 2042-2052CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)The authors present three in silico vol. of distribution at steady state (VDss) models generated on a training set comprising 1096 compds., which goes well beyond the conventional drug space delineated by the Rule of 5 or similar approaches. The authors have performed a careful selection of descriptors and kept a homogeneous Mol. Interaction Field-based descriptor set and linear (Partial Least Squares, PLS) and non-linear (Random Forest, RF) models. The authors have tested the models, which the authors deem orthogonal in nature due to different descriptors and statistical approaches, with good results. In particular the authors tested the RF model, via a leave-class-out approach and by using a set of 34 addnl. compds. not used for training. The authors report comparable results against in vivo scaling approaches with geometric mean-fold error at or below 2 (for a set of 60 compds. with animal data available) and discuss the predictive performance based on the ionization states of the compds. Lastly, the authors report the finding using a two-tier approach (classification and regression) based on VDss ranges, to improve the prediction of compds. with very high VDss. The authors would recommend, overall, the RF model, with 33 descriptors, as the primary choice for VDss prediction in human.
- 4Lombardo, F.; Waters, N. J.; Argikar, U. A.; Dennehy, M. K.; Zhan, J.; Gunduz, M.; Harriman, S. P.; Berellini, G.; Rajlic, I. L.; Obach, R. S. Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 1: volume of distribution at steady state. J. Clin. Pharmacol. 2013, 53, 167– 177, DOI: 10.1177/0091270012440281Google Scholar4https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1yns7fL&md5=446a4a31a0fc877b12f174cdb54c3af1Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 1: volume of distribution at steady stateLombardo, Franco; Waters, Nigel J.; Argikar, Upendra A.; Dennehy, Michelle K.; Zhan, Jenny; Gunduz, Mithat; Harriman, Shawn P.; Berellini, Giuliano; Rajlic, Ivana Liric; Obach, R. ScottJournal of Clinical Pharmacology (2013), 53 (2), 167-177CODEN: JCPCBR; ISSN:0091-2700. (Wiley-Blackwell)The authors present a comprehensive anal. on the estn. of vol. of distribution at steady state (VDss) in human based on rat, dog, and monkey data on nearly 400 compds. for which there are also assocd. human data. This data set, to the authors' knowledge, is the largest publicly available, has been carefully compiled from literature reports, and was expanded with some inhouse detns. such as plasma protein binding data. This work offers a good statistical basis for the evaluation of applicable prediction methods, their accuracy, and some methods-dependent diagnostic tools. The authors also grouped the compds. according to their charge classes and show the applicability of each method considered to each class, offering further insight into the probability of a successful prediction. Furthermore, they found that the use of fraction unbound in plasma, to obtain unbound vol. of distribution, is generally detrimental to accuracy of several methods, and they discuss possible reasons. Overall, the approach using dog and monkey data in the Oie-Tozer equation offers the highest probability of success, with an intrinsic diagnostic tool based on aberrant values (<0 or >1) for the calcd. fraction unbound in tissue. Alternatively, methods based on dog data (single-species scaling) and rat and dog data (Oie-Tozer equation with 2 species or multiple regression methods) may be considered reasonable approaches while not requiring data in nonhuman primates.
- 5Russell, W. M. S.; Burch, R. L. The Principles Of Humane Experimental Technique; Methuen, 1959.Google ScholarThere is no corresponding record for this reference.
- 6Shiran, M. R.; Proctor, N.; Howgate, E.; Rowland-Yeo, K.; Tucker, G.; Rostami-Hodjegan, A. Prediction of metabolic drug clearance in humans: in vitro–in vivo extrapolation vs allometric scaling. Xenobiotica 2006, 36, 567– 580, DOI: 10.1080/00498250600761662Google Scholar6https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28Xpt1eiu7s%253D&md5=eadc44ba7d91a64eabef94c4afc568d5Prediction of metabolic drug clearance in humans: In vitro-in vivo extrapolation vs allometric scalingShiran, M. R.; Proctor, N. J.; Howgate, E. M.; Rowland-Yeo, K.; Tucker, G. T.; Rostami-Hodjegan, A.Xenobiotica (2006), 36 (7), 567-580CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)Previously in vitro-in vivo extrapolation (IVIVE) with the Simcyp Clearance and Interaction Simulator was used to predict the clearance of 15 clin. used drugs in humans. The criteria for the selection of the drugs were that they are used as probes for the activity of specific cytochromes P 450 (CYPs) or have a single CYP isoform as the major or sole contributor to their metab. and that they do not exhibit non-linear kinetics in vivo. Where data were available for the clearance of the drugs in at least 3 animal species, the predictions from IVIVE have now been compared with those based on allometric scaling (AS). Adequate data were available for estg. oral clearance (CLp.o.) in 9 cases (alprazolam, sildenafil, caffeine, clozapine, cyclosporine, dextromethorphan, midazolam, omeprazole, and tolbutamide) and i.v. clearance in 6 cases (CLi.v.) (cyclosporine, diclofenac, midazolam, omeprazole, theophylline, and tolterodine). AS predictions were based on 5 different methods: (1) simple allometry (clearance vs. body wt.); (2) correction for max. life-span potential (CL × MLP); (3) correction for brain wt. (CL × BrW); (4) the use of body surface area; and (5) the rule of exponents. A prediction accuracy was indicated by mean-fold error and the Pearson product moment correlation coeff. Predictions were considered successful if the mean-fold error was ≤2. IVIVE predictions were accurate in 14 of 15 cases (mean-fold error range: 1.02-4.00). All 5 AS methods were accurate in 13, 11, 10, 10, and 14 cases, resp. However, in some cases the error of AS exceeded 5-fold. On the basis of the current results, IVIVE is more reliable than AS in predicting human clearance values for drugs mainly metabolized by CYP450 enzymes. This suggests that the place of AS methods in pre-clin. drug development warrants further scrutiny.
- 7Lombardo, F.; Waters, N. J.; Argikar, U. A.; Dennehy, M. K.; Zhan, J.; Gunduz, M.; Harriman, S. P.; Berellini, G.; Liric Rajlic, I.; Obach, R. S. Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 2: clearance. J. Clin. Pharmacol. 2013, 53, 178– 191, DOI: 10.1177/0091270012440282Google Scholar7https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1ynsL%252FK&md5=a87b8d0a0025910b592a3e757fb35d54Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 2: clearanceLombardo, Franco; Waters, Nigel J.; Argikar, Upendra A.; Dennehy, Michelle K.; Zhan, Jenny; Gunduz, Mithat; Harriman, Shawn P.; Berellini, Giuliano; Rajlic, Ivana Liric; Obach, R. ScottJournal of Clinical Pharmacology (2013), 53 (2), 178-191CODEN: JCPCBR; ISSN:0091-2700. (Wiley-Blackwell)A comprehensive anal. on the prediction of human clearance based on i.v. pharmacokinetic data from rat, dog, and monkey for approx. 400 compds. was undertaken. This data set has been carefully compiled from literature reports and expanded with some inhouse detns. for plasma protein binding and rat clearance. To the authors' knowledge, this is the largest publicly available data set. The present examn. offers a comparison of 37 different methods for prediction of human clearance across compds. of diverse physicochem. properties. Furthermore, this work demonstrates the application of each prediction method to each charge class of the compds., thus presenting an addnl. dimension to prediction of human pharmacokinetics. In general, the observations suggest that methods employing monkey clearance values and a method incorporating differences in plasma protein binding between rat and human yield the best overall predictions as suggested by approx. 60% compds. within 2-fold geometric mean-fold error. Other single-species scaling or proportionality methods incorporating the fraction unbound in the corresponding preclin. species for prediction of free clearance in human were generally unsuccessful.
- 8(a) Crouch, R. D.; Hutzler, J. M.; Daniels, J. S. A novel in vitro allometric scaling methodology for aldehyde oxidase substrates to enable selection of appropriate species for traditional allometry. Xenobiotica 2018, 48, 219– 231, DOI: 10.1080/00498254.2017.1296208Google Scholar8ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXktFyksrs%253D&md5=824ee5fbec02c33e03f72007f16f4b1cA novel in vitro allometric scaling methodology for aldehyde oxidase substrates to enable selection of appropriate species for traditional allometryCrouch, Rachel D.; Hutzler, J. Matthew; Daniels, J. ScottXenobiotica (2018), 48 (3), 219-231CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)1. Failure to predict human pharmacokinetics of aldehyde oxidase (AO) substrates using traditional allometry has been attributed to species differences in AO metab.2. To identify appropriate species for predicting human in vivo clearance by single-species scaling (SSS) or multispecies allometry (MA), we scaled in vitro intrinsic clearance (CLint) of five AO substrates obtained from hepatic S9 of mouse, rat, guinea pig, monkey and minipig to human in vitro CLint.3. When predicting human in vitro CLint, av. abs. fold-error was ≤2.0 by SSS with monkey, minipig and guinea pig (rat/mouse >3.0) and was <3.0 by most MA species combinations (including rat/mouse combinations).4. Interspecies variables, including fraction metabolized by AO (Fm,AO) and hepatic extn. ratios (E) were estd. in vitro. SSS prediction fold-errors correlated with the animal:human ratio of E (r2 = 0.6488), but not Fm,AO (r2 = 0.0051).5. Using plasma clearance (CLp) from the literature, SSS with monkey was superior to rat or mouse at predicting human CLp of BIBX1382 and zoniporide, consistent with in vitro SSS assessments.6. Evaluation of in vitro allometry, Fm,AO and E may prove useful to guide selection of suitable species for traditional allometry and prediction of human pharmacokinetics of AO substrates.(b) Mahmood, I. A Single Animal Species-Based Prediction of Human Clearance and First-in-Human Dose of Monoclonal Antibodies: Beyond Monkey. Antibodies 2021, 10, 35, DOI: 10.3390/antib10030035Google Scholar8bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisleis7bM&md5=8545ead3abf547c1ca6ae9928735162aA Single Animal Species-Based Prediction of Human Clearance and First-in-Human Dose of Monoclonal Antibodies: Beyond MonkeyMahmood, IftekharAntibodies (2021), 10 (3), 35CODEN: ANTICA; ISSN:2073-4468. (MDPI AG)These days, there is a lot of emphasis on the prediction of human clearance (CL) from a single species for monoclonal antibodies (mabs). Many studies indicate that monkey is the most suitable species for the prediction of human clearance for mabs. However, it is not well established if rodents (mouse or rat) can also be used to predict human CL for mabs. The objectives of this study were to predict and compare human CL as well as first-in-human dose of mabs from mouse or rat, ormonkey. Four methods were used for the prediction of human CL of mabs. These methods were: use of four allometric exponents (0.75, 0.80, 0.85, and 0.90), a minimal physiol. based pharmacokinetics method (mPBPK), lymph flow rate, and liver blood flow rate. Based on the predicted CL, first-in-human dose of mabs was projected using either exponent 1.0 (linear scaling) or exponent 0.85, and human-equiv. dose (HED) from each of these species. The results of the study indicated that rat or mouse could provide a reasonably accurate prediction of human CL as well as first-in-human dose of mabs. When exponent 0.85 was used for CL prediction, there were 78%, 95%, and 92% observations within a 2-fold prediction error for mouse, rat, and monkey, resp. Predicted human dose fell within the obsd. human dose range (administered to humans) for 10 out of 13 mabs for mouse, 11 out of 12 mabs for rat, and 12 out of 15 mabs for monkey. Overall, the clearance and first-in-human dose of mabs were predicted reasonably well by all three species (a single species). On av., monkey may be the best species for the prediction of human clearance and human dose but mouse or rat esp.; rat can be a very useful species for conducting the aforementioned studies.(c) Sasabe, H.; Koga, T.; Furukawa, M.; Matsunaga, M.; Kaneko, Y.; Koyama, N.; Hirao, Y.; Akazawa, H.; Kawabata, M.; Kashiyama, E.; Takeuchi, K. Pharmacokinetics and metabolism of brexpiprazole, a novel serotonin-dopamine activity modulator and its main metabolite in rat, monkey and human. Xenobiotica 2021, 51, 590– 604, DOI: 10.1080/00498254.2021.1890275Google Scholar8chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXmtVWnt74%253D&md5=01dd1149ad290f54501e8131d74a17bdPharmacokinetics and metabolism of brexpiprazole, a novel serotonin-dopamine activity modulator and its main metabolite in rat, monkey and humanSasabe, Hiroyuki; Koga, Toshihisa; Furukawa, Masayuki; Matsunaga, Masayuki; Kaneko, Yosuke; Koyama, Noriyuki; Hirao, Yukihiro; Akazawa, Hitomi; Kawabata, Mitsuhiko; Kashiyama, Eiji; Takeuchi, KenjiXenobiotica (2021), 51 (5), 590-604CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)The pharmacokinetics of brexpiprazole were investigated in the in vitro and in vivo. The total body clearance of brexpiprazole in rat and monkey was 2.32 and 0.326 L/h/kg, resp., after i.v. administration, and oral availability was 13.6% and 31.0%, resp. Dose-dependent exposures were obsd. at dose ranges between 1-30 mg/kg in the rat and 0.1-3 mg/kg in the monkey. Brexpiprazole distributed widely to body tissues, and Vd,z were 2.81 and 1.82 L/kg in rat and monkey, resp. The serum protein binding of brexpiprazole was 99% or more in animals and human. Uniform distribution character among the species was suggested by a traditional animal scale-up method. A common main metabolite, DM-3411 was found in animals and humans in the metabolic reactions with the liver S9 fraction. CYP3A4 and CYP2D6 were predominantly involved in the metab. The affinity of DM-3411 for D2 receptors was lower than that of brexpiprazole, and neither DM-3411 nor any metabolites with affinity other than M3 were detected in the brain, demonstrating that brexpiprazole is only involved in the pharmacol. effects. Overall, brexpiprazole has a simple pharmacokinetic profile with good metabolic stability, linear kinetics, and no remarkable species differences with regard to metab. and tissue distribution.
- 9(a) Wang, Y.; Liu, H.; Fan, Y.; Chen, X.; Yang, Y.; Zhu, L.; Zhao, J.; Chen, Y.; Zhang, Y. In Silico Prediction of Human Intravenous Pharmacokinetic Parameters with Improved Accuracy. J. Chem. Inf. Model. 2019, 59, 3968– 3980, DOI: 10.1021/acs.jcim.9b00300Google Scholar9ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhsFKlu7%252FI&md5=84880e17505f8988427afbf981754533In Silico Prediction of Human Intravenous Pharmacokinetic Parameters with Improved AccuracyWang, Yuchen; Liu, Haichun; Fan, Yuanrong; Chen, Xingye; Yang, Yan; Zhu, Lu; Zhao, Junnan; Chen, Yadong; Zhang, YanminJournal of Chemical Information and Modeling (2019), 59 (9), 3968-3980CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Human pharmacokinetics is of great significance in the selection of drug candidates, and in silico estn. of pharmacokinetic parameters in the early stage of drug development has become the trend of drug research owing to its time- and cost-saving advantages. Herein, quant. structure-property relationship studies were carried out to predict four human pharmacokinetic parameters including vol. of distribution at steady state (VDss), clearance (CL), terminal half-life (t1/2), and fraction unbound in plasma (fu), using a data set consisting of 1352 drugs. A series of regression models were built using the most suitable features selected by Boruta algorithm and four machine learning methods including support vector machine (SVM), random forest (RF), gradient boosting machine (GBM), and XGBoost (XGB). For VDss, SVM showed the best performance with R2test = 0.870 and RMSEtest = 0.208. For the other three pharmacokinetic parameters, the RF models produced the superior prediction accuracy (for CL, R2test = 0.875 and RMSEtest = 0.103; for t1/2, R2test = 0.832 and RMSEtest = 0.154; for fu, R2test = 0.818 and RMSEtest = 0.291). Assessed by 10-fold cross validation, leave-one-out cross validation, Y-randomization test and applicability domain evaluation, these models demonstrated excellent stability and predictive ability. Compared with other published models for human pharmacokinetic parameters estn., it was further confirmed that our models obtained better predictive ability and could be used in the selection of preclin. candidates.(b) Gombar, V. K.; Hall, S. D. Quantitative structure-activity relationship models of clinical pharmacokinetics: clearance and volume of distribution. J. Chem. Inf. Model. 2013, 53, 948– 957, DOI: 10.1021/ci400001uGoogle Scholar9bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3sXjt1Wltb0%253D&md5=f7d5b6c91e646c37b3eed26730ac087aQuantitative Structure-Activity Relationship Models of Clinical Pharmacokinetics: Clearance and Volume of DistributionGombar, Vijay K.; Hall, Stephen D.Journal of Chemical Information and Modeling (2013), 53 (4), 948-957CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Reliable prediction of two fundamental human pharmacokinetic (PK) parameters, systemic clearance (CL) and apparent vol. of distribution (Vd), det. the size and frequency of drug dosing and are at the heart of drug discovery and development. Traditionally, estd. CL and Vd are derived from preclin. in vitro and in vivo absorption, distribution, metab., and excretion (ADME) measurements. In this paper, we report quant. structure-activity relationship (QSAR) models for prediction of systemic CL and steady-state Vd (Vdss) from i.v. (iv) dosing in humans. These QSAR models avoid uncertainty assocd. with preclin.-to-clin. extrapolation and require two-dimensional structure drawing as the sole input. The clean, uniform training sets for these models were derived from the compilation published by Obach et al. Models for CL and Vdss were developed using both a support vector regression (SVR) method and a multiple linear regression (MLR) method. The SVR models employ a min. of 2048-bit fingerprints developed inhouse as structure quantifiers. The MLR models, on the other hand, are based on information-rich electro-topol. states of two-atom fragments as descriptors and afford reverse QSAR (RQSAR) anal. to help model-guided, in silico modulation of structures for desired CL and Vdss. The capability of the models to predict iv CL and Vdss with acceptable accuracy was established by randomly splitting data into training and test sets. On av., for both CL and Vdss, 75% of test compds. were predicted within 2.5-fold of the value obsd. and 90% of test compds. were within 5.0-fold of the value obsd. The performance of the final models developed from 525 compds. for CL and 569 compds. for Vdss was evaluated on an external set of 56 compds. The predictions were either better or comparable to those predicted by other in silico models reported in the literature. To demonstrate the practical application of the RQSAR approach, the structure of vildagliptin, a high-CL and a high-Vdss compd., is modified based on the at. contributions to its predicted CL and Vdss to propose compds. with lower CL and lower Vdss.(c) Demir-Kavuk, O.; Bentzien, J.; Muegge, I.; Knapp, E. W. DemQSAR: predicting human volume of distribution and clearance of drugs. J. Comput. Aided Mol. Des. 2011, 25, 1121– 1133, DOI: 10.1007/s10822-011-9496-zGoogle Scholar9chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1emtr3K&md5=5bf5e53dac907c1d63ff4a254eda34daDemQSAR: predicting human volume of distribution and clearance of drugsDemir-Kavuk, Ozgur; Bentzien, Joerg; Muegge, Ingo; Knapp, Ernst-WalterJournal of Computer-Aided Molecular Design (2011), 25 (12), 1121-1133CODEN: JCADEQ; ISSN:0920-654X. (Springer)In silico methods characterizing mol. compds. with respect to pharmacol. relevant properties can accelerate the identification of new drugs and reduce their development costs. Quant. structure-activity/-property relationship (QSAR/QSPR) correlate structure and physico-chem. properties of mol. compds. with a specific functional activity/property under study. Typically a large no. of mol. features are generated for the compds. In many cases the no. of generated features exceeds the no. of mol. compds. with known property values that are available for learning. Machine learning methods tend to overfit the training data in such situations, i.e. the method adjusts to very specific features of the training data, which are not characteristic for the considered property. This problem can be alleviated by diminishing the influence of unimportant, redundant or even misleading features. A better strategy is to eliminate such features completely. Ideally, a mol. property can be described by a small no. of features that are chem. interpretable. The purpose of the present contribution is to provide a predictive modeling approach, which combines feature generation, feature selection, model building and control of overtraining into a single application called DemQSAR. DemQSAR is used to predict human vol. of distribution (VDss) and human clearance (CL). To control overtraining, quadratic and linear regularization terms were employed. A recursive feature selection approach is used to reduce the no. of descriptors. The prediction performance is as good as the best predictions reported in the recent literature. The example presented here demonstrates that DemQSAR can generate a model that uses very few features while maintaining high predictive power. A standalone DemQSAR Java application for model building of any user defined property as well as a web interface for the prediction of human VDss and CL is available on the webpage of DemPRED.
- 10(a) Kosugi, Y.; Hosea, N. Direct Comparison of Total Clearance Prediction: Computational Machine Learning Model versus Bottom-Up Approach Using In Vitro Assay. Mol. Pharm. 2020, 17, 2299– 2309, DOI: 10.1021/acs.molpharmaceut.9b01294Google Scholar10ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVCntbzP&md5=a5278cde7fc6a93ad9ded73f3d01d8aaDirect Comparison of Total Clearance Prediction: Computational Machine Learning Model versus Bottom-Up Approach Using In Vitro AssayKosugi, Yohei; Hosea, NatalieMolecular Pharmaceutics (2020), 17 (7), 2299-2309CODEN: MPOHBP; ISSN:1543-8384. (American Chemical Society)The in vitro-in vivo extrapolation (IVIVE) approach for predicting total plasma clearance (CLtot) has been widely used to rank order compds. early in discovery. More recently, a computational machine learning approach utilizing physicochem. descriptors and fingerprints calcd. from chem. structure information has emerged, enabling virtual predictions even earlier in discovery. Previously, this approach focused more on in vitro intrinsic clearance (CLint) prediction. Herein, we directly compare these two approaches for predicting CLtot in rats. A structurally diverse set of 1114 compds. with known in vivo CLtot, in vitro CLint, and plasma protein binding was used as the basis for this evaluation. The machine learning models were assessed by validation approaches using the time- and cluster-split training and test sets, and five-fold cross validation. Assessed by five-fold validation, the random forest regression (RF) and radial basis function (RBF) models demonstrated better prediction performance in eight attempted machine learning models. The CLtot values predicted by the RF and RBF models were within two-fold of the obsd. values for 67.7 and 71.9% of cluster-split test set compds., resp., while the predictivity was worse in the time-split dataset. The predictivity of both models tended to be improved by incorporating in vitro parameters, unbound fraction in plasma (fu,p), and CLint. CLtot prediction utilizing in vitro CLint and the well-stirred model, correcting for the fraction unbound in blood, was substantially worse compared to machine learning approaches for the same cluster-split test set. The reason that CLtot is underestimated by IVIVE is not fully explained by considering the calcd. microsomal unbound fraction (cfu,mic), extended clearance classification system (ECCS), and omitting high clearance compds. in excess of hepatic blood flow. The anal. suggests that in silico machine learning models may have the power to reduce reliance on or replace in vitro and in vivo studies for chem. structure optimization in early drug discovery.(b) Miljković, F.; Martinsson, A.; Obrezanova, O.; Williamson, B.; Johnson, M.; Sykes, A.; Bender, A.; Greene, N. Machine Learning Models for Human In Vivo Pharmacokinetic Parameters with In-House Validation. Mol. Pharm. 2021, 18, 4520– 4530, DOI: 10.1021/acs.molpharmaceut.1c00718Google Scholar10bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisVGrsLvK&md5=2ee0de42033d3f12164fe12ceff2a63dMachine Learning Models for Human In Vivo Pharmacokinetic Parameters with In-House ValidationMiljkovic, Filip; Martinsson, Anton; Obrezanova, Olga; Williamson, Beth; Johnson, Martin; Sykes, Andy; Bender, Andreas; Greene, NigelMolecular Pharmaceutics (2021), 18 (12), 4520-4530CODEN: MPOHBP; ISSN:1543-8384. (American Chemical Society)Prior to clin. development, a comprehensive pharmacokinetic characterization of a novel drug is required to understand its exposure at the site of action and elimination. Accordingly, in vitro assays and animal pharmacokinetic studies are regularly employed to predict drug exposure in humans, which is often costly and time-consuming. For this reason, the prediction of human pharmacokinetics at the point of design would be of high value for drug discovery. Therefore, we have established a comprehensive data curation protocol that enables machine learning evaluation of 12 human in vivo pharmacokinetic parameters using only chem. structure information and available doses for 1001 unique compds. These machine learning models were thoroughly investigated and validated using both an independent hold-out test set and AstraZeneca clin. data. In addn., the availability of preclin. predictions for a subset of internal clin. candidates allowed us to compare our in silico approach with state-of-the-art pharmacokinetic predictions. Based on this evaluation, three fit-for-purpose models for AUC PO (Rtest2 = 0.63; RMSEtest = 0.76), Cmax PO (Rtest2 = 0.68; RMSEtest = 0.62), and Vdss IV (Rtest2 = 0.47; RMSEtest = 0.50) were identified. Based on the findings, our machine learning models have considerable potential for practical applications in drug discovery, such as influencing decision-making in drug discovery projects and progression of drug candidates toward the clinic.
- 11Iwata, H.; Matsuo, T.; Mamada, H.; Motomura, T.; Matsushita, M.; Fujiwara, T.; Kazuya, M.; Handa, K. Prediction of total drug clearance in humans using animal data: proposal of a multimodal learning method based on deep learning. J. Pharm. Sci. 2021, 110, 1834, DOI: 10.1016/j.xphs.2021.01.020Google Scholar11https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXjs1yrt7w%253D&md5=6be4060a2333878006a36762ecdcd00bPrediction of Total Drug Clearance in Humans Using Animal Data: Proposal of a Multimodal Learning Method Based on Deep LearningIwata, Hiroaki; Matsuo, Tatsuru; Mamada, Hideaki; Motomura, Takahisa; Matsushita, Mayumi; Fujiwara, Takeshi; Kazuya, Maeda; Handa, KoichiJournal of Pharmaceutical Sciences (Philadelphia, PA, United States) (2021), 110 (4), 1834-1841CODEN: JPMSAE; ISSN:0022-3549. (Elsevier Inc.)Research into pharmacokinetics plays an important role in the development process of new drugs. Accurately predicting human pharmacokinetic parameters from preclin. data can increase the success rate of clin. trials. Since clearance (CL) which indicates the capacity of the entire body to process a drug is one of the most important parameters, many methods have been developed. However, there are still rooms to be improved for practical use in drug discovery research; improving CL prediction accuracy and understanding the chem. structure of compds. in terms of pharmacokinetics. To improve those, this research proposes a multimodal learning method based on deep learning that takes not only the chem. structure of a drug but also rat CL as inputs. Good results were obtained compared with the conventional animal scale-up method; the geometric mean fold error was 2.68 and the proportion of compds. with prediction errors of 2-fold or less was 48.5%. Furthermore, it was found to be possible to infer the partial structure useful for CL prediction by a structure contributing factor inference method. The validity of these results of structural interpretation of metabolic stability was confirmed by chemists.
- 12Troyanskaya, O.; Cantor, M.; Sherlock, G.; Brown, P.; Hastie, T.; Tibshirani, R.; Botstein, D.; Altman, R. B. Missing value estimation methods for DNA microarrays. Bioinformatics 2001, 17, 520– 525, DOI: 10.1093/bioinformatics/17.6.520Google Scholar12https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXltFOgsLY%253D&md5=d2c973e7f6355dec54e7fd2c9a599eb1Missing value estimation methods for DNA microarraysTroyanskaya, Olga; Cantor, Michael; Sherlock, Gavin; Brown, Pat; Hastie, Trevor; Tibshirani, Robert; Botstein, David; Altman, Russ B.Bioinformatics (2001), 17 (6), 520-525CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)Motivation: Gene expression microarray expts. can generate data sets with multiple missing expression values. Unfortunately, many algorithms for gene expression anal. require a complete matrix of gene array values as input. For example, methods such as hierarchical clustering and K-means clustering are not robust to missing data, and may lose effectiveness even with a few missing values. Methods for imputing missing data are needed, therefore, to minimize the effect of incomplete data sets on analyses, and to increase the range of data sets to which these algorithms can be applied. In this report, we investigate automated methods for estg. missing data. Results: We present a comparative study of several methods for the estn. of missing values in gene microarray data. We implemented and evaluated three methods: a Singular Value Decompn. (SVD) based method (SVDimpute), weighted K-nearest neighbors (KNNimpute), and row av. We evaluated the methods using a variety of parameter settings and over different real data sets, and assessed the robustness of the imputation methods to the amt. of missing data over the range of 1-20% missing values. We show that KNNimpute appears to provide a more robust and sensitive method for missing value estn. than SVDimpute, and both SVDimpute and KNNimpute surpass the commonly used row av. method (as well as filling missing values with zeros). We report results of the comparative expts. and provide recommendations and tools for accurate estn. of missing microarray data under a variety of conditions.
- 13Schafer, J. L.; Olsen, M. K. Multiple imputation for multivariate missing-data problems: A data analyst’s perspective. Multivariate Behavioral Res. 1998, 33, 545– 571, DOI: 10.1207/s15327906mbr3304_5Google Scholar13https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BC28rpsVOrsw%253D%253D&md5=bdaa91e85de9e3ec3259976d4b62d907Multiple Imputation for Multivariate Missing-Data Problems: A Data Analyst's PerspectiveSchafer J L; Olsen M KMultivariate behavioral research (1998), 33 (4), 545-71 ISSN:0027-3171.Analyses of multivariate data are frequently hampered by missing values. Until recently, the only missing-data methods available to most data analysts have been relatively ad1 hoc practices such as listwise deletion. Recent dramatic advances in theoretical and computational statistics, however, have produced anew generation of flexible procedures with a sound statistical basis. These procedures involve multiple imputation (Rubin, 1987), a simulation technique that replaces each missing datum with a set of m > 1 plausible values. The rn versions of the complete data are analyzed by standard complete-data methods, and the results are combined using simple rules to yield estimates, standard errors, and p-values that formally incorporate missing-data uncertainty. New computational algorithms and software described in a recent book (Schafer, 1997a) allow us to create proper multiple imputations in complex multivariate settings. This article reviews the key ideas of multiple imputation, discusses the software programs currently available, and demonstrates their use on data from the Adolescent Alcohol Prevention Trial (Hansen & Graham, 199 I).
- 14Stekhoven, D. J.; Buhlmann, P. MissForest--non-parametric missing value imputation for mixed-type data. Bioinformatics 2012, 28, 112– 118, DOI: 10.1093/bioinformatics/btr597Google Scholar14https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1yms7fO&md5=44a690989dad7424d50a1662f5de2625MissForest-non-parametric missing value imputation for mixed-type dataStekhoven, Daniel J.; Buehlmann, PeterBioinformatics (2012), 28 (1), 112-118CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)Modern data acquisition based on high-throughput technol. is often facing the problem of missing data. Algorithms commonly used in the anal. of such large-scale data often depend on a complete set. Missing value imputation offers a soln. to this problem. However, the majority of available imputation methods are restricted to one type of variable only: continuous or categorical. For mixed-type data, the different types are usually handled sep. Therefore, these methods ignore possible relations between variable types. We propose a non-parametric method which can cope with different types of variables simultaneously. Results: We compare several state of the art methods for the imputation of missing values. We propose and evaluate an iterative imputation method (missForest) based on a random forest. By averaging over many unpruned classification or regression trees, random forest intrinsically constitutes a multiple imputation scheme. Using the built-in out-of-bag error ests. of random forest, we are able to est. the imputation error without the need of a test set. Evaluation is performed on multiple datasets coming from a diverse selection of biol. fields with artificially introduced missing values ranging from 10% to 30%. We show that missForest can successfully handle missing values, particularly in datasets including different types of variables. In our comparative study, missForest outperforms other methods of imputation esp. in data settings where complex interactions and non-linear relations are suspected. The out-of-bag imputation error ests. of missForest prove to be adequate in all settings. Addnl., missForest exhibits attractive computational efficiency and can cope with high-dimensional data.
- 15Sawada, R.; Iwata, H.; Mizutani, S.; Yamanishi, Y. Target-Based Drug Repositioning Using Large-Scale Chemical-Protein Interactome Data. J. Chem. Inf. Model. 2015, 55, 2717– 2730, DOI: 10.1021/acs.jcim.5b00330Google Scholar15https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2MXhvVGhs73M&md5=5a1ee95bc842228b878d644d692fd798Target-Based Drug Repositioning Using Large-Scale Chemical-Protein Interactome DataSawada, Ryusuke; Iwata, Hiroaki; Mizutani, Sayaka; Yamanishi, YoshihiroJournal of Chemical Information and Modeling (2015), 55 (12), 2717-2730CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Drug repositioning, or the identification of new indications for known drugs, is a useful strategy for drug discovery. In this study, the authors developed novel computational methods to predict potential drug targets and new drug indications for systematic drug repositioning using large-scale chem.-protein interactome data. The authors explored the target space of drugs (including primary targets and off-targets) based on chem. structure similarity and phenotypic effect similarity by making optimal use of millions of compd.-protein interactions. On the basis of the target profiles of drugs, the authors constructed statistical models to predict new drug indications for a wide range of diseases with various mol. features. The proposed method outperformed previous methods in terms of interpretability, applicability, and accuracy. Finally, the authors conducted a comprehensive prediction of the drug-target-disease assocn. network for 8270 drugs and 1401 diseases and showed biol. meaningful examples of newly predicted drug targets and drug indications. The predictive model is useful to understand the mechanisms of the predicted drug indications.
- 16Martin, E. J.; Polyakov, V. R.; Zhu, X. W.; Tian, L.; Mukherjee, P.; Liu, X. All-Assay-Max2 pQSAR: Activity Predictions as Accurate as Four-Concentration IC50s for 8558 Novartis Assays. J. Chem. Inf. Model. 2019, 59, 4450– 4459, DOI: 10.1021/acs.jcim.9b00375Google Scholar16https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhslOqurbK&md5=00bbbad59334754948074d6e1e5edccbAll-Assay-Max2 pQSAR: Activity Predictions as Accurate as Four-Concentration IC50s for 8558 Novartis AssaysMartin, Eric J.; Polyakov, Valery R.; Zhu, Xiang-Wei; Tian, Li; Mukherjee, Prasenjit; Liu, XinJournal of Chemical Information and Modeling (2019), 59 (10), 4450-4459CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Profile-quant. structure-activity relationship (pQSAR) is a massively multitask, two-step machine learning method with unprecedented scope, accuracy, and applicability domain. In step one, a "profile" of conventional single-assay random forest regression models are trained on a very large no. of biochem. and cellular pIC50 assays using Morgan 2 substructural fingerprints as compd. descriptors. In step two, a panel of partial least squares (PLS) models are built using the profile of pIC50 predictions from those random forest regression models as compd. descriptors (hence the name). Previously described for a panel of 728 biochem. and cellular kinase assays, we have now built an enormous pQSAR from 11 805 diverse Novartis (NVS) IC50 and EC50 assays. This large no. of assays, and hence of compd. descriptors for PLS, dictated reducing the profile by only including random forest regression models whose predictions correlate with the assay being modeled. The random forest regression and pQSAR models were evaluated with our "realistically novel" held-out test set, whose median av. similarity to the nearest training set member across the 11 805 assays was only 0.34, comparable to the novelty of compds. actually selected from virtual screens. For the 11 805 single-assay random forest regression models, the median correlation of prediction with the expt. was only rext2 = 0.05, virtually random, and only 8% of the models achieved our std. success threshold of rext2 = 0.30. For pQSAR, the median correlation was rext2 = 0.53, comparable to four-concn. exptl. IC50s, and 72% of the models met our rext2 > 0.30 std., totaling 8558 successful models. The successful models included assays from all of the 51 annotated target subclasses, as well as 4196 phenotypic assays, indicating that pQSAR can be applied to virtually any disease area. Every month, all models are updated to include new measurements, and predictions are made for 5.5 million NVS compds., totaling 50 billion predictions. Common uses have included virtual screening, selectivity design, toxicity and promiscuity prediction, mechanism-of-action prediction, and others. Several such actual applications are described.
- 17Maruhashi, K.; Todoriki, M.; Ohwa, T.; Goto, K.; Hasegawa, Y.; Inakoshi, H.; Anai, H. Learning Multi-way Relations Via Tensor Decomposition with Neural Networks, In Thirty-Second AAAI Conference on Artificial Intelligence , 2018.Google ScholarThere is no corresponding record for this reference.
- 18Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 2012, 40, D1100– D1107, DOI: 10.1093/nar/gkr777Google Scholar18https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs12htbjN&md5=aedf7793e1ca54b6a4fa272ea3ef7d0eChEMBL: a large-scale bioactivity database for drug discoveryGaulton, 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.
- 19Kim, S.; Thiessen, P. A.; Bolton, E. E.; Chen, J.; Fu, G.; Gindulyte, A.; Han, L.; He, J.; He, S.; Shoemaker, B. A.; Wang, J.; Yu, B.; Zhang, J.; Bryant, S. H. PubChem Substance and Compound databases. Nucleic Acids Res. 2016, 44, D1202– 1213, DOI: 10.1093/nar/gkv951Google Scholar19https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXhtV2gu7bE&md5=1ba53f15667506b761d05f0f02313892PubChem substance and compound databasesKim, Sunghwan; Thiessen, Paul A.; Bolton, Evan E.; Chen, Jie; Fu, Gang; Gindulyte, Asta; Han, Lianyi; He, Jane; He, Siqian; Shoemaker, Benjamin A.; Wang, Jiyao; Yu, Bo; Zhang, Jian; Bryant, Stephen H.Nucleic Acids Research (2016), 44 (D1), D1202-D1213CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)PubChem (https://pubchem.ncbi.nlm.nih.gov) is a public repository for information on chem. substances and their biol. activities, launched in 2004 as a component of the Mol. Libraries Roadmap Initiatives of the US National Institutes of Health (NIH). For the past 11 years, PubChem has grown to a sizable system, serving as a chem. information resource for the scientific research community. PubChem consists of three inter-linked databases, Substance, Compd. and BioAssay. The Substance database contains chem. information deposited by individual data contributors to PubChem, and the Compd. database stores unique chem. structures extd. from the Substance database. Biol. activity data of chem. substances tested in assay expts. are contained in the BioAssay database. This paper provides an overview of the PubChem Substance and Compd. databases, including data sources and contents, data organization, data submission using PubChem Upload, chem. structure standardization, web-based interfaces for textual and non-textual searches, and programmatic access. It also gives a brief description of PubChem3D, a resource derived from theor. three-dimensional structures of compds. in PubChem, as well as PubChemRDF, Resource Description Framework (RDF)-formatted PubChem data for data sharing, anal. and integration with information contained in other databases.
- 20Wishart, D. S.; Feunang, Y. D.; Guo, A. C.; Lo, E. J.; Marcu, A.; Grant, J. R.; Sajed, T.; Johnson, D.; Li, C.; Sayeeda, Z.; Assempour, N.; Iynkkaran, I.; Liu, Y.; Maciejewski, A.; Gale, N.; Wilson, A.; Chin, L.; Cummings, R.; Le, D.; Pon, A.; Knox, C.; Wilson, M. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 2018, 46, D1074– D1082, DOI: 10.1093/nar/gkx1037Google Scholar20https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXitlGisbvI&md5=986b28c7ea546596a26dd3ba38f05feeDrugBank 5.0: a major update to the DrugBank database for 2018Wishart, David S.; Feunang, Yannick D.; Guo, An C.; Lo, Elvis J.; Marcu, Ana; Grant, Jason R.; Sajed, Tanvir; Johnson, Daniel; Li, Carin; Sayeeda, Zinat; Assempour, Nazanin; Iynkkaran, Ithayavani; Liu, Yifeng; Maciejewski, Adam; Gale, Nicola; Wilson, Alex; Chin, Lucy; Cummings, Ryan; Le, Diana; Pon, Allison; Knox, Craig; Wilson, MichaelNucleic Acids Research (2018), 46 (D1), D1074-D1082CODEN: NARHAD; ISSN:1362-4962. (Oxford University Press)DrugBank is a web-enabled database contg. comprehensivemol. information about drugs, their mechanisms, their interactions and their targets. First described in 2006, Drug- Bank has continued to evolve over the past 12 years in response to marked improvements to web stds. and changing needs for drug research and development. This year's update, DrugBank 5.0, represents the most significant upgrade to the database in more than 10 years. In many cases, existing data content has grown by 100% or more over the last update. For instance, the total no. of investigational drugs in the database has grown by almost 300%, the no. of drug-drug interactions has grown by nearly 600% and the no. of SNP-assocd. drug effects has grown more than 3000%. Significant improvements have been made to the quantity, quality and consistency of drug indications, drug binding data as well as drug-drug and drug-food interactions. A great deal of brand new data have also been added to DrugBank 5.0. This includes information on the influence of hundreds of drugs on metabolite levels (pharmacometabolomics), gene expression levels (pharmacotranscriptomics) and protein expression levels (pharmacoproteomics). New data have also been added on the status of hundreds of newdrug clin. trials and existing drug repurposing trials. Many other important improvements in the content, interface and performance of the DrugBank website have been made and these should greatly enhance its ease of use, utility and potential applications in many areas of pharmacol. research, pharmaceutical science and drug education.
- 21Varma, M. V. S.; Feng, B.; Obach, R. S.; Troutman, M. D.; Chupka, J.; Miller, H. R.; El-Kattan, A. Physicochemical determinants of human renal clearance. J. Med. Chem. 2009, 52, 4844– 4852, DOI: 10.1021/jm900403jGoogle Scholar21https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXlvFOnt7s%253D&md5=a99ccaa1651d04ed7f7c0a49679f55f6Physicochemical Determinants of Human Renal ClearanceVarma, Manthena V. S.; Feng, Bo; Obach, R. Scott; Troutman, Matthew D.; Chupka, Jonathan; Miller, Howard R.; El-Kattan, AymanJournal of Medicinal Chemistry (2009), 52 (15), 4844-4852CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)Kidney plays an important role in the elimination of drugs, esp. with low or negligible hepatic clearance. An anal. of the interrelation of physicochem. properties and the human renal clearance for a data set of 391 drugs or compds. tested in humans is presented. The data set indicated that lipophilicity shows a neg. relationship while polar descriptors show a pos. relationship with renal clearance. Anal. of net secreted and net reabsorbed subsets revealed that hydrophilic ionized compds. are probable compds. to show net secretion and a possible drug-drug interaction due to their likely interaction with uptake transporters and inherent low passive reabsorption. The physicochem. space and renal clearance were also statistically analyzed by therapeutic area. In conclusion, ionization state, lipophilicity, and polar descriptors are found to be the physicochem. determinants of renal clearance. These fundamental properties can be valuable in early prediction of human renal clearance and can aid the chemist in structural modifications to optimize drug disposition.
- 22Jahandideh-Tehrani, M.; Bozorg-Haddad, O.; Loaiciga, H. A. Application of particle swarm optimization to water management: an introduction and overview. Environ. Monit. Assess. 2020, 192, 281, DOI: 10.1007/s10661-020-8228-zGoogle Scholar22https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB38zkslWqsg%253D%253D&md5=9b05b68e5c836b1fffd6cd2b0069d6f7Application of particle swarm optimization to water management: an introduction and overviewJahandideh-Tehrani Mahsa; Bozorg-Haddad Omid; Loaiciga Hugo AEnvironmental monitoring and assessment (2020), 192 (5), 281 ISSN:.Particle swarm optimization (PSO) is a stochastic population-based optimization algorithm inspired by the interactions of individuals in a social world. This algorithm is widely applied in different fields of water resources problems. This paper presents a comprehensive overview of the basic PSO algorithm search strategy and PSO's applications and performance analysis in water resources engineering optimization problems. Our literature review revealed 22 different varieties of the PSO algorithm. The characteristics of each PSO variety together with their applications in different fields of water resources engineering (e.g., reservoir operation, rainfall-runoff modeling, water quality modeling, and groundwater modeling) are highlighted. The performances of different PSO variants were compared with other evolutionary algorithms (EAs) and mathematical optimization methods. The review evaluates the capability and comparative performance of PSO variants over conventional EAs (e.g., simulated annealing, differential evolution, genetic algorithm, and shark algorithm) and mathematical methods (e.g., support vector machine and differential dynamic programming) in terms of proper convergence to optimal Pareto fronts, faster convergence rate, and diversity of computed solutions.
- 23(a) Breiman, L. Random forests. Mach. Learn. 2001, 45, 5– 32, DOI: 10.1023/A:1010933404324Google ScholarThere is no corresponding record for this reference.(b) Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V. Scikit-learn: Machine learning in Python. J. Machine Learning Res. 2011, 12, 2825– 2830Google ScholarThere is no corresponding record for this reference.
- 24Glorot, X.; Bordes, A.; Bengio, Y. In Deep Sparse Rectifier Neural Networks . Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics2011; pp 315– 323.Google ScholarThere is no corresponding record for this reference.
- 25Ioffe, S.; Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift . arXiv preprint arXiv:1502.03167 2015.Google ScholarThere is no corresponding record for this reference.
- 26Srivastava, N.; Hinton, G.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overfitting. J. Machine Learning Res. 2014, 15, 1929– 1958Google ScholarThere is no corresponding record for this reference.
- 27Varoquaux, G.; Buitinck, L.; Louppe, G.; Grisel, O.; Pedregosa, F.; Mueller, A. Scikit-learn: Machine learning without learning the machinery. GetMobile: Mobile Comput. Commun. 2015, 19, 29– 33, DOI: 10.1145/2786984.2786995Google ScholarThere is no corresponding record for this reference.
- 28Tang, H.; Mayersohn, M. A novel model for prediction of human drug clearance by allometric scaling. Drug Metab. Dispos. 2005, 33, 1297– 1303, DOI: 10.1124/dmd.105.004143Google Scholar28https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXpsFamtrc%253D&md5=4592ad4e0aba395d3755bdefc50209f5A novel model for prediction of human drug clearance by allometric scalingTang, Huadong; Mayersohn, MichaelDrug Metabolism and Disposition (2005), 33 (9), 1297-1303CODEN: DMDSAI; ISSN:0090-9556. (American Society for Pharmacology and Experimental Therapeutics)Sixty-one sets of clearance (CL) values in animal species were allometrically scaled for predicting human clearance. Unbound fractions (fu) of drug in plasma in rats and humans were obtained from the literature. A model was developed to predict human CL: CL = 33.35 mL/min × (a/Rfu)0.770, where Rfu is the fu ratio between rats and humans and a is the coeff. obtained from allometric scaling. The new model was compared with simple allometric scaling and the "rule of exponents" (ROE). Results indicated that the new model provided better predictability for human values of CL than did ROE. It is esp. significant that for the first time the proposed model improves the prediction of CL for drugs illustrating large vertical allometry.
- 29(a) Jones, R. D.; Jones, H. M.; Rowland, M.; Gibson, C. R.; Yates, J. W.; Chien, J. Y.; Ring, B. J.; Adkison, K. K.; Ku, M. S.; He, H. PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 2: comparative assessment of prediction methods of human volume of distribution. J. Pharm. Sci. 2011, 100, 4074– 4089, DOI: 10.1002/jps.22553Google Scholar29ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtVylur%252FE&md5=7c3bf546e194856406b6ee569430c48bPhRMA CPCDC initiative on predictive models of human pharmacokinetics, Part 2: Comparative assessment of prediction methods of human volume of distributionJones, Rhys Do; Jones, Hannah M.; Rowland, Malcolm; Gibson, Christopher R.; Yates, James W. T.; Chien, Jenny Y.; Ring, Barbara J.; Adkison, Kimberly K.; Ku, M. Sherry; He, Handan; Vuppugalla, Ragini; Marathe, Punit; Fischer, Volker; Dutta, Sandeep; Sinha, Vikash K.; Bjoernsson, Thorir; Lave, Thierry; Poulin, PatrickJournal of Pharmaceutical Sciences (2011), 100 (10), 4074-4089CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)The objective of this study was to evaluate the performance of various empirical, semimechanistic and mechanistic methodologies with and without protein binding corrections for the prediction of human vol. of distribution at steady state (Vss). PhRMA member companies contributed a set of blinded data from preclin. and clin. studies, and 18 drugs with i.v. clin. pharmacokinetics (PK) data were available for the anal. In vivo and in vitro preclin. data were used to predict Vss by 24 different methods. Various statistical and outlier techniques were employed to assess the predictability of each method. There was not simply one method that predicts Vss accurately for all compds. Across methods, the max. success rate in predicting human Vss was 100%, 94%, and 78% of the compds. with predictions falling within tenfold, threefold, and twofold error, resp., of the obsd. Vss. Generally, the methods that made use of in vivo preclin. data were more predictive than those methods that relied solely on in vitro data. However, for many compds., in vivo data from only two species (generally rat and dog) were available and/or the required in vitro data were missing, which meant some methods could not be properly evaluated. It is recommended to initially use the in vitro tissue compn.-based equations to predict Vss in preclin. species and humans, putting the assumptions and compd. properties into context. As in vivo data become available, these predictions should be reassessed and rationalized to indicate the level of confidence (uncertainty) in the human Vss prediction. The top three methods that perform strongly at integrating in vivo data in this way were the Oie-Tozer, the rat -dog-human proportionality equation, and the lumped-PBPK approach. Overall, the scientific benefit of this study was to obtain greater characterization of predictions of human Vss from several methods available in the literature. © 2011 Wiley-Liss, Inc. and the American Pharmacists Assocn. J Pharm Sci 100:4074-4089, 2011.(b) Obach, R. S.; Baxter, J. G.; Liston, T. E.; Silber, B. M.; Jones, B. C.; Macintyre, F.; Rance, D. J.; Wastall, P. The prediction of human pharmacokinetic parameters from preclinical and in vitro metabolism data. J. Pharmacol. Exp. Ther. 1997, 283, 46– 58Google Scholar29bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2sXmslCltr0%253D&md5=68d1910d925e26ec4a8ef23043ffb1edThe prediction of human pharmacokinetic parameters from preclinical and in vitro metabolism dataObach, R. Scott; Baxter, James G.; Liston, Theodore E.; Silber, B. Michael; Jones, Barry C.; Macintyre, Flona; Rance, David J.; Wastall, PhilipJournal of Pharmacology and Experimental Therapeutics (1997), 283 (1), 46-58CODEN: JPETAB; ISSN:0022-3565. (Williams & Wilkins)A review with 35 refs. We describe a comprehensive retrospective anal. in which the abilities of several methods by which human pharmacokinetic parameters are predicted from preclin. pharmacokinetic data and/or in vitro metab. data were assessed. The prediction methods examd. included both methods from the scientific literature as well as some described in this report for the first time. Four methods were examd. for their ability to predict human vol. of distribution. Three were highly predictive, yielding, on av., predictions that were within 60% to 90% of actual values. Twelve methods were assessed for their utility in predicting clearance. The most successful allometric scaling method yielded clearance predictions that were, on av., within 80% of actual values. The best methods in which in vitro metab. data from human liver microsomes were scaled to in vivo clearance values yielded predicted clearance values that were, on av., within 70% to 80% of actual values. Human t1/2 was predicted by combining predictions of human vol. of distribution and clearance. The best t1/2 prediction methods successfully assigned compds. to appropriate dosing regimen categories (e.g., once daily, twice daily and so forth) 70% to 80% of the time. In addn., correlations between human t1/2 and t1/2 values from preclin. species were also generally successful (72-87%) when used to predict human dosing regimens. In summary, this retrospective anal. has identified several approaches by which human pharmacokinetic data can be predicted from preclin. data. Such approaches should find utility in the drug discovery and development processes in the identification and selection of compds. that will possess appropriate pharmacokinetic characteristics in humans for progression to clin. trials.
- 30Słoczyńska, K.; Gunia-Krzyzak, A.; Koczurkiewicz, P.; Wojcik-Pszczola, K.; Zelaszczyk, D.; Popiol, J.; Pekala, E. Metabolic stability and its role in the discovery of new chemical entities. Acta. Pharm. 2019, 69, 345– 361, DOI: 10.2478/acph-2019-0024Google Scholar30https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXntFWhtg%253D%253D&md5=2dfad07056bce4e4cc32264c8ce07040Metabolic stability and its role in the discovery of new chemical entitiesSloczynska, Karolina; Gunia-Krzyzak, Agnieszka; Koczurkiewicz, Paulina; Wojcik-Pszczola, Katarzyna; Zelaszczyk, Dorota; Popiol, Justyna; Pekala, ElzbietaActa Pharmaceutica (Warsaw, Poland) (2019), 69 (3), 345-361CODEN: ACPHEE; ISSN:1846-9558. (Sciendo)A review. Detn. of metabolic profiles of new chem. entities is a key step in the process of drug discovery, since it influences pharmacokinetic characteristics of therapeutic compds. One of the main challenges of medicinal chem. is not only to design compds. demonstrating beneficial activity, but also mols. exhibiting favorable pharmacokinetic parameters. Chem. compds. can be divided into those which are metabolized relatively fast and those which undergo slow biotransformation. Rapid biotransformation reduces exposure to the maternal compd. and may lead to the generation of active, non-active or toxic metabolites. In contrast, high metabolic stability may promote interactions between drugs and lead to parent compd. toxicity. In the present paper, issues of compd. metabolic stability will be discussed, with special emphasis on its significance, in vitro metabolic stability testing, dilemmas regarding in vitro-in vivo extrapolation of the results and some aspects relating to different preclin. species used in in vitro metabolic stability assessment of compds.
- 31Wakayama, N.; Toshimoto, K.; Maeda, K.; Hotta, S.; Ishida, T.; Akiyama, Y.; Sugiyama, Y. In Silico Prediction of Major Clearance Pathways of Drugs among 9 Routes with Two-Step Support Vector Machines. Pharm. Res. 2018, 35, 197, DOI: 10.1007/s11095-018-2479-1Google Scholar31https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB3c3hs12mtw%253D%253D&md5=94b5e6b45c2eb894cdcc4abe0d7b0f81In Silico Prediction of Major Clearance Pathways of Drugs among 9 Routes with Two-Step Support Vector MachinesWakayama Naomi; Toshimoto Kota; Hotta Shun; Ishida Takashi; Akiyama Yutaka; Toshimoto Kota; Sugiyama Yuichi; Maeda KazuyaPharmaceutical research (2018), 35 (10), 197 ISSN:.PURPOSE: The clearance pathways of drugs are critical elements for understanding the pharmacokinetics of drugs. We previously developed in silico systems to predict the five clearance pathway using a rectangular method and a support vector machine (SVM). In this study, we improved our classification system by increasing the number of clearance pathways available for our prediction (CYP1A2, CYP2C8, CYP2C19, and UDP-glucuronosyl transferases (UGTs)) and by accepting multiple major pathways. METHODS: Using the four default descriptors (charge, molecular weight, logD at pH 7.0, and unbound fraction in plasma), three kinds of SVM-based predictors based on traditional single-step approach or two-step focusing approaches with subset or partition clustering were developed. The two-step approach with subset clustering resulted in the highest prediction performance. The feature-selection of additional descriptors based on a greedy algorithm was employed to further improve the predictability. RESULTS: The prediction accuracy for each pathway was increased to more than 0.83 with the exception of CYP2C19 and UGTs pathways, whose accuracies were below 0.7. Prediction performance of CYP1A2, CYP3A4 and renal excretion pathways were found to be acceptable using external dataset. CONCLUSIONS: We successfully constructed a novel SVM-based predictor for the multiple major clearance pathways based on chemical structures.
- 32Poulin, P.; Dambach, D. M.; Hartley, D. H.; Ford, K.; Theil, F. P.; Harstad, E.; Halladay, J.; Choo, E.; Boggs, J.; Liederer, B. M.; Dean, B.; Diaz, D. An algorithm for evaluating potential tissue drug distribution in toxicology studies from readily available pharmacokinetic parameters. J. Pharm. Sci. 2013, 102, 3816– 3829, DOI: 10.1002/jps.23670Google Scholar32https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3sXhtFCitLvM&md5=61c57bce0b889c7dab87369a8b186b0eAn Algorithm for Evaluating PotentKpial Tissue Drug Distribution in Toxicology Studies from Readily Available Pharmacokinetic ParametersPoulin, Patrick; Dambach, Donna M.; Hartley, Dylan H.; Ford, Kevin; Theil, Frank-Peter; Harstad, Eric; Halladay, Jason; Choo, Edna; Boggs, Jason; Liederer, Bianca M.; Dean, Brian; Diaz, DoloresJournal of Pharmaceutical Sciences (2013), 102 (10), 3816-3829CODEN: JPMSAE; ISSN:0022-3549. (John Wiley & Sons, Inc.)Having an understanding of drug tissue accumulation can be informative in the assessment of target organ toxicities; however, obtaining tissue drug levels from toxicol. studies by bioanal. methods is labor-intensive and infrequently performed. Addnl., there are no described methods for predicting tissue drug distribution for the exptl. conditions in toxicol. studies, which typically include non-steady-state conditions and very high exposures that may sat. several processes. The aim was the development of an algorithm to provide semiquant. and quant. ests. of tissue-to-plasma concn. ratios (Kp) for several tissues from readily available parameters of pharmacokinetics (PK) such as vol. of distribution (Vd) and clearance of each drug, without performing tissue measurement in vivo. The computational approach is specific for the oral route of administration and non-steady-state conditions and was applied for a dataset of 29 Genentech small mols. such as neutral compds. as well as weak and strong org. bases. The max. success rate in predicting Kp values within 2.5-fold error of obsd. Kp values was 82% at low doses (<100 mg/kg) in preclin. species. Prediction accuracy was relatively lower with satn. at high doses (≥100 mg/kg); however, an approach to perform low-to-high dose extrapolations of Kp values was presented and applied successfully in most cases. An approach for the interspecies scaling was also applied successfully. Finally, the proposed algorithm was used in a case study and successfully predicted differential tissue distribution of two small-mol. MET kinase inhibitors, which had different toxicity profiles in mice. This newly developed algorithm can be used to predict the partition coeffs. Kp for small mols. in toxicol. studies, which can be leveraged to optimize the PK drivers of tissue distribution in an attempt to decrease drug tissue level, and improve safety margins. © 2013 Wiley Periodicals, Inc. and the American Pharmacists Assocn. J Pharm Sci.
- 33(a) Poulin, P.; Theil, F. P. A priori prediction of tissue: plasma partition coefficients of drugs to facilitate the use of physiologically-based pharmacokinetic models in drug discovery. J. Pharm. Sci. 2000, 89, 16– 35, DOI: 10.1002/(SICI)1520-6017(200001)89:1<16::AID-JPS3>3.0.CO;2-EGoogle Scholar33ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXit1eis7c%253D&md5=c72666192591b0768bec8aaed1608cf8A priori prediction of tissue:plasma partition coefficients of drugs to facilitate the use of physiologically-based pharmacokinetic models in drug discoveryPoulin, Patrick; Theil, Frank-PeterJournal of Pharmaceutical Sciences (2000), 89 (1), 16-35CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)The tissue:plasma (Pt:p) partition coeffs. (PCs) are important drug-specific input parameters in physiol. based pharmacokinetic (PBPK) models used to est. the disposition of drugs in biota. Until now the use of PBPK models in early stages of the drug discovery process was not possible, since the estn. of Pt:p of new drug candidates by using conventional in vitro and/or in vivo methods is too time and cost intensive. The objectives of the study were (i) to develop and validate two mechanistic equations for predicting a priori the rabbit, rat and mouse Pt:p of non-adipose and non-excretory tissues (bone, brain, heart, intestine, lung, muscle, skin, spleen) for 65 structurally unrelated drugs and (ii) to evaluate the adequacy of using Pt:p of muscle as predictors for Pt:p of other tissues. The first equation predicts Pt:p at steady state, assuming a homogenous distribution and passive diffusion of drugs in tissues, from a ratio of soly. and macromol. binding between tissues and plasma. The ratio of soly. was estd. from log vegetable oil:water PCs (Kvo:w) of drugs and lipid and water levels in tissues and plasma, whereas the ratio of macromol. binding for drugs was estd. from tissue interstitial fluid-to-plasma concn. ratios of albumin, globulins and lipoproteins. The second equation predicts Pt:p of drugs residing predominantly in the interstitial space of tissues. Therefore, the fractional vol. content of interstitial space in each tissue replaced drug solubilities in the first equation. Following the development of these equations, regression analyses between Pt:p of muscle and those of the other tissues were examd. The av. ratio of predicted-to-exptl. Pt:p values was 1.26 (SD = 1.40, r = 0.90), and 85% of the 269 predicted values were within a factor of three of the corresponding literature values obtained under in vivo and in vitro conditions. For predicted and exptl. Pt:p, linear relationships (r > 0.9 in most cases) were obsd. between muscle and other tissues, suggesting that Pt:p of muscle is a good predictor for the Pt:p of other tissues. The two previous equations could explain the mechanistic basis of these linear relationships. The practical aim of this study is a worthwhile goal for pharmacokinetic screening of new drug candidates.(b) Poulin, P.; Schoenlein, K.; Theil, F. P. Prediction of adipose tissue: plasma partition coefficients for structurally unrelated drugs. J. Pharm. Sci. 2001, 90, 436– 447, DOI: 10.1002/1520-6017(200104)90:4<436::AID-JPS1002>3.0.CO;2-PGoogle Scholar33bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXivVant7s%253D&md5=c9af9c58e95ca183fdd30f94ea367edbPrediction of adipose tissue:plasma partition coefficients for structurally unrelated drugsPoulin, Patrick; Schoenlein, Kerstin; Theil, Frank-PeterJournal of Pharmaceutical Sciences (2001), 90 (4), 436-447CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)Tissue:plasma (Pt:p) partition coeffs. (PCs) are important parameters describing tissue distribution of drugs. The ultimate goal in early drug discovery is to develop and validate in silico methods for predicting a priori the Pt:p for each new drug candidate. In this context, tissue compn.-based equations have recently been developed and validated for predicting a priori the non-adipose and adipose Pt:p for neutral org. solvents and pollutants. For ionizable drugs that bind to different degrees to common plasma proteins, only their non-adipose Pt:p values have been predicted with these equations. The only compd.-dependent input parameters for these equations are the lipophilicity parameter, such as olive oil-water PC (Kvo:w) or n-octanol-water PC (Po:w), and/or unbound fraction in plasma (fup) detd. under in vitro conditions. Tissue compn.-based equations could potentially also be used to predict adipose tissue-plasma PCs (Pat:p) for ionized drugs. The main objective of the present study was to modify these equations for predicting in vivo Pat:p (white fat) for 14 structurally unrelated ionized drugs that bind substantially to plasma macromols. in rats, rabbits, or humans. The second objective was to verify whether Kvo:w or Po:w provides more accurate predictions of in vivo Pat:p (i.e., to verify whether olive oil or n-octanol is the better surrogate for lipids in adipose tissue). The second objective was supported by comparing in vitro data on Pat:p with those on olive oil-plasma PC (Kvo:p) for five drugs. Furthermore, in vivo Pat:p was not only predicted from Kvo:w and Po:w of the non-ionized species, but also from Kvo:w* and Po:w*, taking into account the ionized species in addn. The Pat:p predicted from Kvo:w*, Po:w*, and Po:w differ from the in vivo Pat:p by an av. factor of 1.17 (SD = 0.44, r = 0.95), 15.0 (SD = 15.7, r = 0.59), and 40.7 (SD = 57.2, r = 0.33), resp. The in vitro values of Kvo:p differ from those of Pat:p by an av. factor of 0.86 (SD = 0.16, r = 0.99, n = 5). The results demonstrate that (i) the equation using only data on fup as input and olive oil as lipophilicity surrogate is able to provide accurate predictions of in vivo Pat:p, and (ii) olive oil is a better surrogate of the adipose tissue lipids than n-octanol. The present study is an innovative method for predicting in vivo fat partitioning of drugs in mammals.(c) Poulin, P.; Krishnan, K. A biologically-based algorithm for predicting human tissue: blood partition coefficients of organic chemicals. Hum. Exp. Toxicol. 1995, 14, 273– 280, DOI: 10.1177/096032719501400307Google Scholar33chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2MXls1ersL0%253D&md5=131ccbfb41ef67bbf75ea2fcfd774e56A biologically-based algorithm for predicting human tissue: blood partition coefficients of organic chemicalsPoulin, Patrick; Krishnan, KannanHuman & Experimental Toxicology (1995), 14 (3), 273-80CODEN: HETOEA; ISSN:0960-3271.A biol.-based algorithm for predicting the tissue: blood partition coeffs. (PCs) of org. chems. has been developed. The approach consisted of (i) describing tissues and blood in terms of their neutral lipid, phospholipid, and water contents, (ii) obtaining data on the soly. of chems. in n-octanol and water, and (iii) calcg. the tissue: blood PCs by assuming that the soly. of a chem. in n-octanol corresponds to its soly. in neutral lipids, the soly. in water corresponds to the soly. in tissue/blood water fraction, and the soly. in phospholipids is a function of soly. in water and n-octanol. The adequacy of this approach was verified by comparing the predicted values with previously published exptl. data on human tissue (liver, lung, muscle, kidney, brain, adipose tissue): blood PCs for 23 org. chems. In the case of liver, lung, and muscle, the predicted PC values were in close agreement with the higher-end of the range of exptl. PC values found in the literature. The predicted brain: and kidney: blood PCs were greater than the exptl. PCs in most cases by approx. a factor of two. Whereas the adipose tissue: blood PCs of relatively less hydrophilic chems. were adequately predicted, the predicted PCs for relatively more hydrophilic chems. were much greater than the exptl.-detd. values. There was a good agreement between the predicted and exptl.-detd. blood soly. of the 23 chems. chosen for this study, indicating that the overestn. of tissue:blood PCs by the present method is not due to under-estn. of blood soly. of chems. Rather, it might be due to the lower tissue soly. of chems. obsd. exptl. due to the complexity of the tissue matrixes. This novel approach of describing tissues in terms of the type of lipid and water content should enable the prediction of the tissue:blood PCs of org. chems. with information on their soly. in water and n-octanol, for developing physiol.-based toxicokinetic models.(d) Berezhkovskiy, L. M. Volume of distribution at steady state for a linear pharmacokinetic system with peripheral elimination. J. Pharm. Sci. 2004, 93, 1628– 1640, DOI: 10.1002/jps.20073Google Scholar33dhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXks12hsrY%253D&md5=55af00dd6f68f348a617cbef9784757dVolume of distribution at steady state for a linear pharmacokinetic system with peripheral eliminationBerezhkovskiy, Leonid M.Journal of Pharmaceutical Sciences (2004), 93 (6), 1628-1640CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)A review. The problem of finding the steady-state vol. of distribution Vss for a linear pharmacokinetic system with peripheral drug elimination is considered. A commonly used equation Vss=(D/AUC)*MRT is applicable only for the systems with central (plasma) drug elimination. The following equation, Vss=(D/AUC)*MRTint, was obtained, where AUC is the commonly calcd. area under the time curve of the total drug concn. in plasma after i.v. administration of bolus drug dose, D, and MRTint is the intrinsic mean residence time, which is the av. time the drug spends in the body (system) after entering the systemic circulation (plasma). The value of MRTint cannot be found from a drug plasma concn. profile after an i.v. bolus drug input if a peripheral drug exit occurs. The obtained equation does not contain the assumption of an immediate equil. of protein and tissue binding in plasma and organs, and thus incorporates the rates of all possible reactions. If drug exits the system only through central compartment (plasma) and there is an instant equil. between bound and unbound drug fractions in plasma, then MRTint becomes equal to MRT=AUMC/AUC, which is calcd. using the time course of the total drug concn. in plasma after an i.v. bolus injection. Thus, the obtained equation coincides with the traditional one, Vss=(D/AUC)*MRT, if the assumptions for validity of this equation are met. Exptl. methods for detg. the steady-state vol. of distribution and MRTint, as well as the problem of detg. whether peripheral drug elimination occurs, are considered. The equation for calcn. of the tissue-plasma partition coeff. with the account of peripheral elimination is obtained. The difference between traditionally calcd. Vss=(D/AUC)*MRT and the true value given by (D/AUC)*MRTint is discussed.(e) Rodgers, T.; Rowland, M. Physiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterions. J. Pharm. Sci. 2006, 95, 1238– 1257, DOI: 10.1002/jps.20502Google Scholar33ehttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28XlsFWlsrY%253D&md5=6691b676c3ac213a96c8724576ad761aPhysiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterionsRodgers, Trudy; Rowland, MalcolmJournal of Pharmaceutical Sciences (2006), 95 (6), 1238-1257CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)A key component of whole-body physiol. based pharmacokinetic (WBPBPK) models is the tissue-to-plasma water partition coeffs. (Kpu's). The predictability of Kpu values using mechanistically derived equations has been investigated for 7 very weak bases, 20 acids, 4 neutral drugs and 8 zwitterions in rat adipose, bone, brain, gut, heart, kidney, liver, lung, muscle, pancreas, skin, spleen and thymus. These equations incorporate expressions for dissoln. in tissue water and for partitioning into neutral lipids and neutral phospholipids. Addnl., assocns. with acidic phospholipids were incorporated for zwitterions with a highly basic functionality or extracellular proteins for the other compd. classes. The affinity for these cellular constituents was detd. from blood cell data or plasma protein binding, resp. These equations assume drugs are passively distributed and that processes are nonsatg. Resultant Kpu predictions were more accurate when compared to published equations, with 84% as opposed to 61% of the predicted values agreeing with exptl. values to within a factor of 3. This improvement was largely due to the incorporation of distribution processes related to drug ionization, an issue that was not addressed in earlier equations. Such advancements in parameter prediction will assist WBPBPK modeling, where time, cost, and labor requirements greatly deter its application.(f) Schmitt, W. General approach for the calculation of tissue to plasma partition coefficients. Toxicol. In Vitro 2008, 22, 457– 467, DOI: 10.1016/j.tiv.2007.09.010Google Scholar33fhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXhsFygt7o%253D&md5=b2cdca7ad30692ad2ff57ffcbe86a775General approach for the calculation of tissue to plasma partition coefficientsSchmitt, WalterToxicology in Vitro (2008), 22 (2), 457-467CODEN: TIVIEQ; ISSN:0887-2333. (Elsevier Ltd.)A new mechanistic, universal model for the calcn. of steady state tissue:plasma partition coeffs. (Kt:p) of org. chems. in mammalian species was developed. The approach allows the estn. of Kt:p-values based on the compn. of the tissues in terms of water, neutral lipids, neutral and acidic phospholipids and proteins using the lipophilicity, the binding to phospholipid membranes, the pKa and the unbound fraction in blood plasma as compd. specific parameters. Taking explicitly into account the sign and fraction of the charge of the compds. at the physiol. pH the method is universally applicable to neutral, acidic, basic or multiply charged substances and has thus a significantly extended applicability compared to previously published approaches. The model was applied to 59 chem. diverse drug compds. for which tissue:plasma partition coeffs. are reported in the literature. In total 474 exptl. obsd. Kt:p values for 12 tissues and the red blood cells were available and could be compared to model results. For 73% of the calcd. values a deviation less than 3-fold from the resp. obsd. value was found, proving the validity of the approach.
- 34Tess, D. A.; Eng, H.; Kalgutkar, A. S.; Litchfield, J.; Edmonds, D. J.; Griffith, D. A.; Varma, M. V. S. Predicting the Human Hepatic Clearance of Acidic and Zwitterionic Drugs. J. Med. Chem. 2020, 63, 11831– 11844, DOI: 10.1021/acs.jmedchem.0c01033Google Scholar34https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhvFartbvN&md5=3e9b035730014972fa070f7cc4cd6404Predicting the Human Hepatic Clearance of Acidic and Zwitterionic DrugsTess, David A.; Eng, Heather; Kalgutkar, Amit S.; Litchfield, John; Edmonds, David J.; Griffith, David A.; Varma, Manthena V. S.Journal of Medicinal Chemistry (2020), 63 (20), 11831-11844CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)Prospective predictions of human hepatic clearance for anionic/zwitterionic compds., which are oftentimes subjected to transporter-mediated uptake, are challenging in drug discovery. We evaluated the utility of preclin. species, rats and cynomolgus monkeys [nonhuman primates (NHPs)], to predict the human hepatic clearance using a diverse set of acidic/zwitterionic drugs. Preclin. clearance data were generated following i.v. dosing in rats/NHPs and compared to the human clearance data (n = 18/27). Single-species scaling of NHP clearance with an allometric exponent of 0.50 allowed for good prediction of human clearance (fold error ~ 2.1, bias ~ 1.0), with ~ 86% predictions within 3-fold. In comparison, rats underpredicted the clearance of lipophilic acids, while overprediction was noted for hydrophilic acids. Finally, an in vitro clearance assay based on human hepatocytes, which is routinely used in discovery setting, markedly underpredicted human clearance (bias ~ 0.12). Collectively, this study provides insights into the usefulness of the preclin. models in enabling pharmacokinetic optimization for acid/zwitterionic drug candidates.
- 35Miyamoto, M.; Iwasaki, S.; Chisaki, I.; Nakagawa, S.; Amano, N.; Hirabayashi, H. Comparison of predictability for human pharmacokinetics parameters among monkeys, rats, and chimeric mice with humanised liver. Xenobiotica 2017, 47, 1052– 1063, DOI: 10.1080/00498254.2016.1265160Google Scholar35https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXjs1Gksb0%253D&md5=442cba86ccdbe16bb08cee7ef572730bComparison of predictability for human pharmacokinetics parameters among monkeys, rats, and chimeric mice with humanised liverMiyamoto, Maki; Iwasaki, Shinji; Chisaki, Ikumi; Nakagawa, Sayaka; Amano, Nobuyuki; Hirabayashi, HidekiXenobiotica (2017), 47 (12), 1052-1063CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)The aim of the present study was to evaluate the usefulness of chimeric mice with humanised liver (PXB mice) for the prediction of clearance (CLt) and vol. of distribution at steady state (Vdss), in comparison with monkeys, which have been reported as a reliable model for human pharmacokinetics (PK) prediction, and with rats, as a conventional PK model. CLt and Vdss values in PXB mice, monkeys and rats were detd. following i.v. administration of 30 compds. known to be mainly eliminated in humans via the hepatic metab. by various drug-metabolizing enzymes. Using single-species allometric scaling, human CLt and Vdss values were predicted from the three animal models. Predicted CLt values from PXB mice exhibited the highest predictability: 25 for PXB mice, 21 for monkeys and 14 for rats were predicted within a three-fold range of actual values among 30 compds. For predicted human Vdss values, the no. of compds. falling within a three-fold range was 23 for PXB mice, 24 for monkeys, and 16 for rats among 29 compds. PXB mice indicated a higher predictability for CLt and Vdss values than the other animal models. These results demonstrate the utility of PXB mice in predicting human PK parameters.
- 36Russell, W.; Burch, R. The Principles of Humane Experimental Technique. Wheathampstead, Universities Federation for Animal Welfare: UK; 1959.Google ScholarThere is no corresponding record for this reference.
Cited By
Smart citations by scite.ai include citation statements extracted from the full text of the citing article. The number of the statements may be higher than the number of citations provided by ACS Publications if one paper cites another multiple times or lower if scite has not yet processed some of the citing articles.
This article is cited by 24 publications.
- Yuanyuan Zhang, Zhiyin Xie, Fu Xiao, Jie Yu, Zhehuan Fan, Shihui Sun, Jiangshan Shi, Zunyun Fu, Xutong Li, Dingyan Wang, Mingyue Zheng, Xiaomin Luo. Prediction of Multi-Pharmacokinetics Property in Multi-Species: Bayesian Neural Network Stacking Model with Uncertainty. Molecular Pharmaceutics 2024, 21
(12)
, 6177-6192. https://doi.org/10.1021/acs.molpharmaceut.4c00406
- Leonid Komissarov, Nenad Manevski, Katrin Groebke Zbinden, Torsten Schindler, Marinka Zitnik, Lisa Sach-Peltason. Actionable Predictions of Human Pharmacokinetics at the Drug Design Stage. Molecular Pharmaceutics 2024, 21
(9)
, 4356-4371. https://doi.org/10.1021/acs.molpharmaceut.4c00311
- Matthew Adrian, Yunsie Chung, Alan C. Cheng. Denoising Drug Discovery Data for Improved Absorption, Distribution, Metabolism, Excretion, and Toxicity Property Prediction. Journal of Chemical Information and Modeling 2024, 64
(16)
, 6324-6337. https://doi.org/10.1021/acs.jcim.4c00639
- Koichi Handa, Saki Yoshimura, Michiharu Kageyama, Takeshi Iijima. Development of Novel Methods for QSAR Modeling by Machine Learning Repeatedly: A Case Study on Drug Distribution to Each Tissue. Journal of Chemical Information and Modeling 2024, 64
(9)
, 3662-3669. https://doi.org/10.1021/acs.jcim.4c00046
- Franco Lombardo, Jörg Bentzien, Giuliano Berellini, Ingo Muegge. Prediction of Human Clearance Using In Silico Models with Reduced Bias. Molecular Pharmaceutics 2024, 21
(3)
, 1192-1203. https://doi.org/10.1021/acs.molpharmaceut.3c00812
- Yaguo Gong, Wei Ding, Panpan Wang, Qibiao Wu, Xiaojun Yao, Qingxia Yang. Evaluating Machine Learning Methods of Analyzing Multiclass Metabolomics. Journal of Chemical Information and Modeling 2023, 63
(24)
, 7628-7641. https://doi.org/10.1021/acs.jcim.3c01525
- Christopher E. Keefer, George Chang, Li Di, Nathaniel A. Woody, David A. Tess, Sarah M. Osgood, Brendon Kapinos, Jill Racich, Anthony A. Carlo, Amanda Balesano, Nicholas Ferguson, Christine Orozco, Larisa Zueva, Lina Luo. The Comparison of Machine Learning and Mechanistic In Vitro–In Vivo Extrapolation Models for the Prediction of Human Intrinsic Clearance. Molecular Pharmaceutics 2023, 20
(11)
, 5616-5630. https://doi.org/10.1021/acs.molpharmaceut.3c00502
- Koichi Handa, Peter Wright, Saki Yoshimura, Michiharu Kageyama, Takeshi Iijima, Andreas Bender. Prediction of Compound Plasma Concentration–Time Profiles in Mice Using Random Forest. Molecular Pharmaceutics 2023, 20
(6)
, 3060-3072. https://doi.org/10.1021/acs.molpharmaceut.3c00071
- Ion Brinza, Razvan Stefan Boiangiu, Iasmina Honceriu, Ahmed M. Abd-Alkhalek, Samir M. Osman, Omayma A. Eldahshan, Elena Todirascu-Ciornea, Gabriela Dumitru, Lucian Hritcu. Neuroprotective Potential of Origanum majorana L. Essential Oil Against Scopolamine-Induced Memory Deficits and Oxidative Stress in a Zebrafish Model. Biomolecules 2025, 15
(1)
, 138. https://doi.org/10.3390/biom15010138
- Hiroaki Iwata. Transforming drug discovery: the impact of AI and molecular simulation on R&D efficiency. Bioanalysis 2024, 16
(23-24)
, 1211-1217. https://doi.org/10.1080/17576180.2024.2437283
- Jae-Hee Kwon, Ja-Young Han, Minjung Kim, Seong Kyung Kim, Dong-Kyu Lee, Myeong Gyu Kim. Prediction of human pharmacokinetic parameters incorporating SMILES information. Archives of Pharmacal Research 2024, 47
(12)
, 914-923. https://doi.org/10.1007/s12272-024-01520-2
- Ion Brinza, Razvan Stefan Boiangiu, Marius Mihasan, Dragos Lucian Gorgan, Alexandru Bogdan Stache, Ahmed Abd-Alkhalek, Heba El-Nashar, Iriny Ayoub, Nada Mostafa, Omayma Eldahshan, Abdel Nasser Singab, Lucian Hritcu. Rhoifolin, baicalein 5,6-dimethyl ether and agathisflavone prevent amnesia induced in scopolamine zebrafish (Danio rerio) model by increasing the mRNA expression of bdnf, npy, egr-1, nfr2α, and creb1 genes. European Journal of Pharmacology 2024, 984 , 177013. https://doi.org/10.1016/j.ejphar.2024.177013
- Miaoran Ning, Ma Fang, Kushal Shah, Vaishali Dixit, Devendra Pade, Helen Musther, Sibylle Neuhoff. A cross-species assessment of in silico prediction methods of steady-state volume of distribution using Simcyp simulators. Journal of Pharmaceutical Sciences 2024, 32 https://doi.org/10.1016/j.xphs.2024.12.018
- Xiaohua Lu, Liangxu Xie, Lei Xu, Rongzhi Mao, Xiaojun Xu, Shan Chang. Multimodal fused deep learning for drug property prediction: Integrating chemical language and molecular graph. Computational and Structural Biotechnology Journal 2024, 23 , 1666-1679. https://doi.org/10.1016/j.csbj.2024.04.030
- Houria Bentoumi, Abdeslem Bouzina, Aïcha Amira, Omar Sekiou, Djawhara Chohra, Loubna Ferchichi, Rachida Zerrouki, Nour-Eddine Aouf. Theoretical investigations of some isolated compounds from
Calophyllum flavoramulum
as potential antioxidant agents and inhibitors of AGEs. Journal of Biomolecular Structure and Dynamics 2024, 7 , 1-27. https://doi.org/10.1080/07391102.2024.2428375
- Dennis A Smith, Lucy Melanie Burton, Sophie Amanda Smith. Through a computer monitor darkly: artificial intelligence in absorption, distribution, metabolism and excretion science. Xenobiotica 2024, 54
(7)
, 359-367. https://doi.org/10.1080/00498254.2023.2295361
- Davide Bassani, Neil John Parrott, Nenad Manevski, Jitao David Zhang. Another string to your bow: machine learning prediction of the pharmacokinetic properties of small molecules. Expert Opinion on Drug Discovery 2024, 19
(6)
, 683-698. https://doi.org/10.1080/17460441.2024.2348157
- Mahnaz Ahmadi, Bahareh Alizadeh, Seyed Mohammad Ayyoubzadeh, Mahdiye Abiyarghamsari. Predicting Pharmacokinetics of Drugs Using Artificial Intelligence Tools: A Systematic Review. European Journal of Drug Metabolism and Pharmacokinetics 2024, 49
(3)
, 249-262. https://doi.org/10.1007/s13318-024-00883-7
- Hiroaki Iwata, Yoshihiro Hayashi, Takuto Koyama, Aki Hasegawa, Kosuke Ohgi, Ippei Kobayashi, Yasushi Okuno. Feature extraction of particle morphologies of pharmaceutical excipients from scanning electron microscope images using convolutional neural networks. International Journal of Pharmaceutics 2024, 653 , 123873. https://doi.org/10.1016/j.ijpharm.2024.123873
- Hiroshi Komura, Reiko Watanabe, Kenji Mizuguchi. The Trends and Future Prospective of In Silico Models from the Viewpoint of ADME Evaluation in Drug Discovery. Pharmaceutics 2023, 15
(11)
, 2619. https://doi.org/10.3390/pharmaceutics15112619
- Koichi Handa, Sakae Sugiyama, Michiharu Kageyama, Takeshi Iijima. Combined data-driven and mechanism-based approaches for human-intestinal-absorption prediction in the early drug-discovery stage. Digital Discovery 2023, 2
(5)
, 1577-1588. https://doi.org/10.1039/D3DD00144J
- Koichi Handa, Seishiro Sakamoto, Michiharu Kageyama, Takeshi Iijima. Development of a 2D-QSAR Model for Tissue-to-Plasma Partition Coefficient Value with High Accuracy Using Machine Learning Method, Minimum Required Experimental Values, and Physicochemical Descriptors. European Journal of Drug Metabolism and Pharmacokinetics 2023, 48
(4)
, 341-352. https://doi.org/10.1007/s13318-023-00832-w
- Hiroaki Iwata. Application of in Silico Technologies for Drug Target Discovery and Pharmacokinetic Analysis. Chemical and Pharmaceutical Bulletin 2023, 71
(6)
, 398-405. https://doi.org/10.1248/cpb.c22-00638
- Olga Obrezanova. Artificial intelligence for compound pharmacokinetics prediction. Current Opinion in Structural Biology 2023, 79 , 102546. https://doi.org/10.1016/j.sbi.2023.102546
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.
Recommended Articles
Abstract
Figure 1
Figure 1. Workflow of our novel human CLtot and Vdss prediction method. (A) CLtot analysis flow. (i) There were 741 compounds with human CLtot data and 46 that had values for all 11 features. (ii) All feature values were estimated via prediction using ADMEWORKS. (iii) Feature extraction was performed using XGBoost or Random Forest, and a prediction model was constructed. (B) Vdss analysis flow. (i) There were 751 compounds with human Vdss data and 46 that had values for all 11 features. (ii) All feature values were estimated via prediction using ADMEWORKS. (iii) Feature extraction was performed using XGBoost or Random Forest, and a prediction model was constructed.
Figure 2
Figure 2. Overview of the multimodal Deep Tensor model.
References
This article references 36 other publications.
- 1Ballard, P.; Brassil, P.; Bui, K. H.; Dolgos, H.; Petersson, C.; Tunek, A.; Webborn, P. J. The right compound in the right assay at the right time: an integrated discovery DMPK strategy. Drug Metab. Rev. 2012, 44, 224– 252, DOI: 10.3109/03602532.2012.6910991https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC38XhtVelsLzF&md5=105fbbcfba6213caaadf41c8b10cf408The right compound in the right assay at the right time: an integrated discovery DMPK strategyBallard, Peter; Brassil, Patrick; Bui, Khanh H.; Dolgos, Hugues; Petersson, Carl; Tunek, Anders; Webborn, Peter J. H.Drug Metabolism Reviews (2012), 44 (3), 224-252CODEN: DMTRAR; ISSN:0360-2532. (Informa Healthcare)A review. The high rate of attrition during drug development and its assocd. high research and development (R&D) cost have put pressure on pharmaceutical companies to ensure that candidate drugs going to clin. testing have the appropriate quality such that the biol. hypothesis could be evaluated. To help achieve this ambition, drug metab. and pharmacokinetic (DMPK) science and increasing investment have been deployed earlier in the R&D process. To gain max. return on investment, it is essential that DMPK concepts are both appropriately integrated into the compd. design process and that compd. selection is focused on accurate prediction of likely outcomes in patients. This article describes key principles that underpin the contribution of DMPK science for small-mol. research based on 15 years of discovery support in a major pharmaceutical company. It does not aim to describe the breadth and depth of DMPK science, but more the practical application for decision making in real-world situations.
- 2Andrade, E. L.; Bento, A. F.; Cavalli, J.; Oliveira, S. K.; Schwanke, R. C.; Siqueira, J. M.; Freitas, C. S.; Marcon, R.; Calixto, J. B. Non-clinical studies in the process of new drug development - Part II: Good laboratory practice, metabolism, pharmacokinetics, safety and dose translation to clinical studies. Braz. J. Med. Biol. Res. 2016, 49, e5646 DOI: 10.1590/1414-431x201656462https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXhsFarsbg%253D&md5=c76a3d9eb5b8844a77f26a705062578fNon-clinical studies in the process of new drug development - part II: good laboratory practice, metabolism, pharmacokinetics, safety and dose translation to clinical studiesAndrade, E. L.; Bento, A. F.; Cavalli, J.; Oliveir, S. K.; Schwank, R. C.; Siqueir, J. M.; Freita, C. S.; Marcon, R.; Calixto, J. B.Brazilian Journal of Medical and Biological Research (2016), 49 (12), e5646/1-e5646/19CODEN: BJMRDK; ISSN:0100-879X. (Associacao Brasileira de Divulgacao Cientifica)The process of drug development involves non-clin. and clin. studies. Non-clin. studies are conducted using different protocols including animal studies, which mostly follow the Good Lab. Practice (GLP) regulations. During the early preclin. development process, also known as Go/No-Go decision, a drug candidate needs to pass through several steps, such as detn. of drug availability (studies on pharmacokinetics), absorption, distribution, metab. and elimination (ADME) and preliminary studies that aim to investigate the candidate safety including genotoxicity, mutagenicity, safety pharmacol. and general toxicol. These preliminary studies generally do not need to comply with GLP regulations. These studies aim at investigating the drug safety to obtain the first information about its tolerability in different systems that are relevant for further decisions. There are, however, other studies that should be performed according to GLP stds. and are mandatory for the safe exposure to humans, such as repeated dose toxicity, genotoxicity and safety pharmacol. These studies must be conducted before the Investigational New Drug (IND) application. The package of non-clin. studies should cover all information needed for the safe transposition of drugs from animals to humans, generally based on the non-obsd. adverse effect level (NOAEL) obtained from general toxicity studies. After IND approval, other GLP expts. for the evaluation of chronic toxicity, reproductive and developmental toxicity, carcinogenicity and genotoxicity, are carried out during the clin. phase of development. However, the necessity of performing such studies depends on the new drug clin. application purpose.
- 3Lombardo, F.; Jing, Y. In Silico Prediction of Volume of Distribution in Humans. Extensive Data Set and the Exploration of Linear and Nonlinear Methods Coupled with Molecular Interaction Fields Descriptors. J. Chem. Inf. Model. 2016, 56, 2042– 2052, DOI: 10.1021/acs.jcim.6b000443https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC28XhsVOrtLzI&md5=b99a25fcd384ecc777f9190f2061e49aIn Silico Prediction of Volume of Distribution in Humans. Extensive Data Set and the Exploration of Linear and Nonlinear Methods Coupled with Molecular Interaction Fields DescriptorsLombardo, Franco; Jing, YankangJournal of Chemical Information and Modeling (2016), 56 (10), 2042-2052CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)The authors present three in silico vol. of distribution at steady state (VDss) models generated on a training set comprising 1096 compds., which goes well beyond the conventional drug space delineated by the Rule of 5 or similar approaches. The authors have performed a careful selection of descriptors and kept a homogeneous Mol. Interaction Field-based descriptor set and linear (Partial Least Squares, PLS) and non-linear (Random Forest, RF) models. The authors have tested the models, which the authors deem orthogonal in nature due to different descriptors and statistical approaches, with good results. In particular the authors tested the RF model, via a leave-class-out approach and by using a set of 34 addnl. compds. not used for training. The authors report comparable results against in vivo scaling approaches with geometric mean-fold error at or below 2 (for a set of 60 compds. with animal data available) and discuss the predictive performance based on the ionization states of the compds. Lastly, the authors report the finding using a two-tier approach (classification and regression) based on VDss ranges, to improve the prediction of compds. with very high VDss. The authors would recommend, overall, the RF model, with 33 descriptors, as the primary choice for VDss prediction in human.
- 4Lombardo, F.; Waters, N. J.; Argikar, U. A.; Dennehy, M. K.; Zhan, J.; Gunduz, M.; Harriman, S. P.; Berellini, G.; Rajlic, I. L.; Obach, R. S. Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 1: volume of distribution at steady state. J. Clin. Pharmacol. 2013, 53, 167– 177, DOI: 10.1177/00912700124402814https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1yns7fL&md5=446a4a31a0fc877b12f174cdb54c3af1Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 1: volume of distribution at steady stateLombardo, Franco; Waters, Nigel J.; Argikar, Upendra A.; Dennehy, Michelle K.; Zhan, Jenny; Gunduz, Mithat; Harriman, Shawn P.; Berellini, Giuliano; Rajlic, Ivana Liric; Obach, R. ScottJournal of Clinical Pharmacology (2013), 53 (2), 167-177CODEN: JCPCBR; ISSN:0091-2700. (Wiley-Blackwell)The authors present a comprehensive anal. on the estn. of vol. of distribution at steady state (VDss) in human based on rat, dog, and monkey data on nearly 400 compds. for which there are also assocd. human data. This data set, to the authors' knowledge, is the largest publicly available, has been carefully compiled from literature reports, and was expanded with some inhouse detns. such as plasma protein binding data. This work offers a good statistical basis for the evaluation of applicable prediction methods, their accuracy, and some methods-dependent diagnostic tools. The authors also grouped the compds. according to their charge classes and show the applicability of each method considered to each class, offering further insight into the probability of a successful prediction. Furthermore, they found that the use of fraction unbound in plasma, to obtain unbound vol. of distribution, is generally detrimental to accuracy of several methods, and they discuss possible reasons. Overall, the approach using dog and monkey data in the Oie-Tozer equation offers the highest probability of success, with an intrinsic diagnostic tool based on aberrant values (<0 or >1) for the calcd. fraction unbound in tissue. Alternatively, methods based on dog data (single-species scaling) and rat and dog data (Oie-Tozer equation with 2 species or multiple regression methods) may be considered reasonable approaches while not requiring data in nonhuman primates.
- 5Russell, W. M. S.; Burch, R. L. The Principles Of Humane Experimental Technique; Methuen, 1959.There is no corresponding record for this reference.
- 6Shiran, M. R.; Proctor, N.; Howgate, E.; Rowland-Yeo, K.; Tucker, G.; Rostami-Hodjegan, A. Prediction of metabolic drug clearance in humans: in vitro–in vivo extrapolation vs allometric scaling. Xenobiotica 2006, 36, 567– 580, DOI: 10.1080/004982506007616626https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28Xpt1eiu7s%253D&md5=eadc44ba7d91a64eabef94c4afc568d5Prediction of metabolic drug clearance in humans: In vitro-in vivo extrapolation vs allometric scalingShiran, M. R.; Proctor, N. J.; Howgate, E. M.; Rowland-Yeo, K.; Tucker, G. T.; Rostami-Hodjegan, A.Xenobiotica (2006), 36 (7), 567-580CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)Previously in vitro-in vivo extrapolation (IVIVE) with the Simcyp Clearance and Interaction Simulator was used to predict the clearance of 15 clin. used drugs in humans. The criteria for the selection of the drugs were that they are used as probes for the activity of specific cytochromes P 450 (CYPs) or have a single CYP isoform as the major or sole contributor to their metab. and that they do not exhibit non-linear kinetics in vivo. Where data were available for the clearance of the drugs in at least 3 animal species, the predictions from IVIVE have now been compared with those based on allometric scaling (AS). Adequate data were available for estg. oral clearance (CLp.o.) in 9 cases (alprazolam, sildenafil, caffeine, clozapine, cyclosporine, dextromethorphan, midazolam, omeprazole, and tolbutamide) and i.v. clearance in 6 cases (CLi.v.) (cyclosporine, diclofenac, midazolam, omeprazole, theophylline, and tolterodine). AS predictions were based on 5 different methods: (1) simple allometry (clearance vs. body wt.); (2) correction for max. life-span potential (CL × MLP); (3) correction for brain wt. (CL × BrW); (4) the use of body surface area; and (5) the rule of exponents. A prediction accuracy was indicated by mean-fold error and the Pearson product moment correlation coeff. Predictions were considered successful if the mean-fold error was ≤2. IVIVE predictions were accurate in 14 of 15 cases (mean-fold error range: 1.02-4.00). All 5 AS methods were accurate in 13, 11, 10, 10, and 14 cases, resp. However, in some cases the error of AS exceeded 5-fold. On the basis of the current results, IVIVE is more reliable than AS in predicting human clearance values for drugs mainly metabolized by CYP450 enzymes. This suggests that the place of AS methods in pre-clin. drug development warrants further scrutiny.
- 7Lombardo, F.; Waters, N. J.; Argikar, U. A.; Dennehy, M. K.; Zhan, J.; Gunduz, M.; Harriman, S. P.; Berellini, G.; Liric Rajlic, I.; Obach, R. S. Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 2: clearance. J. Clin. Pharmacol. 2013, 53, 178– 191, DOI: 10.1177/00912700124402827https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2cXhs1ynsL%252FK&md5=a87b8d0a0025910b592a3e757fb35d54Comprehensive assessment of human pharmacokinetic prediction based on in vivo animal pharmacokinetic data, part 2: clearanceLombardo, Franco; Waters, Nigel J.; Argikar, Upendra A.; Dennehy, Michelle K.; Zhan, Jenny; Gunduz, Mithat; Harriman, Shawn P.; Berellini, Giuliano; Rajlic, Ivana Liric; Obach, R. ScottJournal of Clinical Pharmacology (2013), 53 (2), 178-191CODEN: JCPCBR; ISSN:0091-2700. (Wiley-Blackwell)A comprehensive anal. on the prediction of human clearance based on i.v. pharmacokinetic data from rat, dog, and monkey for approx. 400 compds. was undertaken. This data set has been carefully compiled from literature reports and expanded with some inhouse detns. for plasma protein binding and rat clearance. To the authors' knowledge, this is the largest publicly available data set. The present examn. offers a comparison of 37 different methods for prediction of human clearance across compds. of diverse physicochem. properties. Furthermore, this work demonstrates the application of each prediction method to each charge class of the compds., thus presenting an addnl. dimension to prediction of human pharmacokinetics. In general, the observations suggest that methods employing monkey clearance values and a method incorporating differences in plasma protein binding between rat and human yield the best overall predictions as suggested by approx. 60% compds. within 2-fold geometric mean-fold error. Other single-species scaling or proportionality methods incorporating the fraction unbound in the corresponding preclin. species for prediction of free clearance in human were generally unsuccessful.
- 8(a) Crouch, R. D.; Hutzler, J. M.; Daniels, J. S. A novel in vitro allometric scaling methodology for aldehyde oxidase substrates to enable selection of appropriate species for traditional allometry. Xenobiotica 2018, 48, 219– 231, DOI: 10.1080/00498254.2017.12962088ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXktFyksrs%253D&md5=824ee5fbec02c33e03f72007f16f4b1cA novel in vitro allometric scaling methodology for aldehyde oxidase substrates to enable selection of appropriate species for traditional allometryCrouch, Rachel D.; Hutzler, J. Matthew; Daniels, J. ScottXenobiotica (2018), 48 (3), 219-231CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)1. Failure to predict human pharmacokinetics of aldehyde oxidase (AO) substrates using traditional allometry has been attributed to species differences in AO metab.2. To identify appropriate species for predicting human in vivo clearance by single-species scaling (SSS) or multispecies allometry (MA), we scaled in vitro intrinsic clearance (CLint) of five AO substrates obtained from hepatic S9 of mouse, rat, guinea pig, monkey and minipig to human in vitro CLint.3. When predicting human in vitro CLint, av. abs. fold-error was ≤2.0 by SSS with monkey, minipig and guinea pig (rat/mouse >3.0) and was <3.0 by most MA species combinations (including rat/mouse combinations).4. Interspecies variables, including fraction metabolized by AO (Fm,AO) and hepatic extn. ratios (E) were estd. in vitro. SSS prediction fold-errors correlated with the animal:human ratio of E (r2 = 0.6488), but not Fm,AO (r2 = 0.0051).5. Using plasma clearance (CLp) from the literature, SSS with monkey was superior to rat or mouse at predicting human CLp of BIBX1382 and zoniporide, consistent with in vitro SSS assessments.6. Evaluation of in vitro allometry, Fm,AO and E may prove useful to guide selection of suitable species for traditional allometry and prediction of human pharmacokinetics of AO substrates.(b) Mahmood, I. A Single Animal Species-Based Prediction of Human Clearance and First-in-Human Dose of Monoclonal Antibodies: Beyond Monkey. Antibodies 2021, 10, 35, DOI: 10.3390/antib100300358bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisleis7bM&md5=8545ead3abf547c1ca6ae9928735162aA Single Animal Species-Based Prediction of Human Clearance and First-in-Human Dose of Monoclonal Antibodies: Beyond MonkeyMahmood, IftekharAntibodies (2021), 10 (3), 35CODEN: ANTICA; ISSN:2073-4468. (MDPI AG)These days, there is a lot of emphasis on the prediction of human clearance (CL) from a single species for monoclonal antibodies (mabs). Many studies indicate that monkey is the most suitable species for the prediction of human clearance for mabs. However, it is not well established if rodents (mouse or rat) can also be used to predict human CL for mabs. The objectives of this study were to predict and compare human CL as well as first-in-human dose of mabs from mouse or rat, ormonkey. Four methods were used for the prediction of human CL of mabs. These methods were: use of four allometric exponents (0.75, 0.80, 0.85, and 0.90), a minimal physiol. based pharmacokinetics method (mPBPK), lymph flow rate, and liver blood flow rate. Based on the predicted CL, first-in-human dose of mabs was projected using either exponent 1.0 (linear scaling) or exponent 0.85, and human-equiv. dose (HED) from each of these species. The results of the study indicated that rat or mouse could provide a reasonably accurate prediction of human CL as well as first-in-human dose of mabs. When exponent 0.85 was used for CL prediction, there were 78%, 95%, and 92% observations within a 2-fold prediction error for mouse, rat, and monkey, resp. Predicted human dose fell within the obsd. human dose range (administered to humans) for 10 out of 13 mabs for mouse, 11 out of 12 mabs for rat, and 12 out of 15 mabs for monkey. Overall, the clearance and first-in-human dose of mabs were predicted reasonably well by all three species (a single species). On av., monkey may be the best species for the prediction of human clearance and human dose but mouse or rat esp.; rat can be a very useful species for conducting the aforementioned studies.(c) Sasabe, H.; Koga, T.; Furukawa, M.; Matsunaga, M.; Kaneko, Y.; Koyama, N.; Hirao, Y.; Akazawa, H.; Kawabata, M.; Kashiyama, E.; Takeuchi, K. Pharmacokinetics and metabolism of brexpiprazole, a novel serotonin-dopamine activity modulator and its main metabolite in rat, monkey and human. Xenobiotica 2021, 51, 590– 604, DOI: 10.1080/00498254.2021.18902758chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXmtVWnt74%253D&md5=01dd1149ad290f54501e8131d74a17bdPharmacokinetics and metabolism of brexpiprazole, a novel serotonin-dopamine activity modulator and its main metabolite in rat, monkey and humanSasabe, Hiroyuki; Koga, Toshihisa; Furukawa, Masayuki; Matsunaga, Masayuki; Kaneko, Yosuke; Koyama, Noriyuki; Hirao, Yukihiro; Akazawa, Hitomi; Kawabata, Mitsuhiko; Kashiyama, Eiji; Takeuchi, KenjiXenobiotica (2021), 51 (5), 590-604CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)The pharmacokinetics of brexpiprazole were investigated in the in vitro and in vivo. The total body clearance of brexpiprazole in rat and monkey was 2.32 and 0.326 L/h/kg, resp., after i.v. administration, and oral availability was 13.6% and 31.0%, resp. Dose-dependent exposures were obsd. at dose ranges between 1-30 mg/kg in the rat and 0.1-3 mg/kg in the monkey. Brexpiprazole distributed widely to body tissues, and Vd,z were 2.81 and 1.82 L/kg in rat and monkey, resp. The serum protein binding of brexpiprazole was 99% or more in animals and human. Uniform distribution character among the species was suggested by a traditional animal scale-up method. A common main metabolite, DM-3411 was found in animals and humans in the metabolic reactions with the liver S9 fraction. CYP3A4 and CYP2D6 were predominantly involved in the metab. The affinity of DM-3411 for D2 receptors was lower than that of brexpiprazole, and neither DM-3411 nor any metabolites with affinity other than M3 were detected in the brain, demonstrating that brexpiprazole is only involved in the pharmacol. effects. Overall, brexpiprazole has a simple pharmacokinetic profile with good metabolic stability, linear kinetics, and no remarkable species differences with regard to metab. and tissue distribution.
- 9(a) Wang, Y.; Liu, H.; Fan, Y.; Chen, X.; Yang, Y.; Zhu, L.; Zhao, J.; Chen, Y.; Zhang, Y. In Silico Prediction of Human Intravenous Pharmacokinetic Parameters with Improved Accuracy. J. Chem. Inf. Model. 2019, 59, 3968– 3980, DOI: 10.1021/acs.jcim.9b003009ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhsFKlu7%252FI&md5=84880e17505f8988427afbf981754533In Silico Prediction of Human Intravenous Pharmacokinetic Parameters with Improved AccuracyWang, Yuchen; Liu, Haichun; Fan, Yuanrong; Chen, Xingye; Yang, Yan; Zhu, Lu; Zhao, Junnan; Chen, Yadong; Zhang, YanminJournal of Chemical Information and Modeling (2019), 59 (9), 3968-3980CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Human pharmacokinetics is of great significance in the selection of drug candidates, and in silico estn. of pharmacokinetic parameters in the early stage of drug development has become the trend of drug research owing to its time- and cost-saving advantages. Herein, quant. structure-property relationship studies were carried out to predict four human pharmacokinetic parameters including vol. of distribution at steady state (VDss), clearance (CL), terminal half-life (t1/2), and fraction unbound in plasma (fu), using a data set consisting of 1352 drugs. A series of regression models were built using the most suitable features selected by Boruta algorithm and four machine learning methods including support vector machine (SVM), random forest (RF), gradient boosting machine (GBM), and XGBoost (XGB). For VDss, SVM showed the best performance with R2test = 0.870 and RMSEtest = 0.208. For the other three pharmacokinetic parameters, the RF models produced the superior prediction accuracy (for CL, R2test = 0.875 and RMSEtest = 0.103; for t1/2, R2test = 0.832 and RMSEtest = 0.154; for fu, R2test = 0.818 and RMSEtest = 0.291). Assessed by 10-fold cross validation, leave-one-out cross validation, Y-randomization test and applicability domain evaluation, these models demonstrated excellent stability and predictive ability. Compared with other published models for human pharmacokinetic parameters estn., it was further confirmed that our models obtained better predictive ability and could be used in the selection of preclin. candidates.(b) Gombar, V. K.; Hall, S. D. Quantitative structure-activity relationship models of clinical pharmacokinetics: clearance and volume of distribution. J. Chem. Inf. Model. 2013, 53, 948– 957, DOI: 10.1021/ci400001u9bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3sXjt1Wltb0%253D&md5=f7d5b6c91e646c37b3eed26730ac087aQuantitative Structure-Activity Relationship Models of Clinical Pharmacokinetics: Clearance and Volume of DistributionGombar, Vijay K.; Hall, Stephen D.Journal of Chemical Information and Modeling (2013), 53 (4), 948-957CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Reliable prediction of two fundamental human pharmacokinetic (PK) parameters, systemic clearance (CL) and apparent vol. of distribution (Vd), det. the size and frequency of drug dosing and are at the heart of drug discovery and development. Traditionally, estd. CL and Vd are derived from preclin. in vitro and in vivo absorption, distribution, metab., and excretion (ADME) measurements. In this paper, we report quant. structure-activity relationship (QSAR) models for prediction of systemic CL and steady-state Vd (Vdss) from i.v. (iv) dosing in humans. These QSAR models avoid uncertainty assocd. with preclin.-to-clin. extrapolation and require two-dimensional structure drawing as the sole input. The clean, uniform training sets for these models were derived from the compilation published by Obach et al. Models for CL and Vdss were developed using both a support vector regression (SVR) method and a multiple linear regression (MLR) method. The SVR models employ a min. of 2048-bit fingerprints developed inhouse as structure quantifiers. The MLR models, on the other hand, are based on information-rich electro-topol. states of two-atom fragments as descriptors and afford reverse QSAR (RQSAR) anal. to help model-guided, in silico modulation of structures for desired CL and Vdss. The capability of the models to predict iv CL and Vdss with acceptable accuracy was established by randomly splitting data into training and test sets. On av., for both CL and Vdss, 75% of test compds. were predicted within 2.5-fold of the value obsd. and 90% of test compds. were within 5.0-fold of the value obsd. The performance of the final models developed from 525 compds. for CL and 569 compds. for Vdss was evaluated on an external set of 56 compds. The predictions were either better or comparable to those predicted by other in silico models reported in the literature. To demonstrate the practical application of the RQSAR approach, the structure of vildagliptin, a high-CL and a high-Vdss compd., is modified based on the at. contributions to its predicted CL and Vdss to propose compds. with lower CL and lower Vdss.(c) Demir-Kavuk, O.; Bentzien, J.; Muegge, I.; Knapp, E. W. DemQSAR: predicting human volume of distribution and clearance of drugs. J. Comput. Aided Mol. Des. 2011, 25, 1121– 1133, DOI: 10.1007/s10822-011-9496-z9chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1emtr3K&md5=5bf5e53dac907c1d63ff4a254eda34daDemQSAR: predicting human volume of distribution and clearance of drugsDemir-Kavuk, Ozgur; Bentzien, Joerg; Muegge, Ingo; Knapp, Ernst-WalterJournal of Computer-Aided Molecular Design (2011), 25 (12), 1121-1133CODEN: JCADEQ; ISSN:0920-654X. (Springer)In silico methods characterizing mol. compds. with respect to pharmacol. relevant properties can accelerate the identification of new drugs and reduce their development costs. Quant. structure-activity/-property relationship (QSAR/QSPR) correlate structure and physico-chem. properties of mol. compds. with a specific functional activity/property under study. Typically a large no. of mol. features are generated for the compds. In many cases the no. of generated features exceeds the no. of mol. compds. with known property values that are available for learning. Machine learning methods tend to overfit the training data in such situations, i.e. the method adjusts to very specific features of the training data, which are not characteristic for the considered property. This problem can be alleviated by diminishing the influence of unimportant, redundant or even misleading features. A better strategy is to eliminate such features completely. Ideally, a mol. property can be described by a small no. of features that are chem. interpretable. The purpose of the present contribution is to provide a predictive modeling approach, which combines feature generation, feature selection, model building and control of overtraining into a single application called DemQSAR. DemQSAR is used to predict human vol. of distribution (VDss) and human clearance (CL). To control overtraining, quadratic and linear regularization terms were employed. A recursive feature selection approach is used to reduce the no. of descriptors. The prediction performance is as good as the best predictions reported in the recent literature. The example presented here demonstrates that DemQSAR can generate a model that uses very few features while maintaining high predictive power. A standalone DemQSAR Java application for model building of any user defined property as well as a web interface for the prediction of human VDss and CL is available on the webpage of DemPRED.
- 10(a) Kosugi, Y.; Hosea, N. Direct Comparison of Total Clearance Prediction: Computational Machine Learning Model versus Bottom-Up Approach Using In Vitro Assay. Mol. Pharm. 2020, 17, 2299– 2309, DOI: 10.1021/acs.molpharmaceut.9b0129410ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhtVCntbzP&md5=a5278cde7fc6a93ad9ded73f3d01d8aaDirect Comparison of Total Clearance Prediction: Computational Machine Learning Model versus Bottom-Up Approach Using In Vitro AssayKosugi, Yohei; Hosea, NatalieMolecular Pharmaceutics (2020), 17 (7), 2299-2309CODEN: MPOHBP; ISSN:1543-8384. (American Chemical Society)The in vitro-in vivo extrapolation (IVIVE) approach for predicting total plasma clearance (CLtot) has been widely used to rank order compds. early in discovery. More recently, a computational machine learning approach utilizing physicochem. descriptors and fingerprints calcd. from chem. structure information has emerged, enabling virtual predictions even earlier in discovery. Previously, this approach focused more on in vitro intrinsic clearance (CLint) prediction. Herein, we directly compare these two approaches for predicting CLtot in rats. A structurally diverse set of 1114 compds. with known in vivo CLtot, in vitro CLint, and plasma protein binding was used as the basis for this evaluation. The machine learning models were assessed by validation approaches using the time- and cluster-split training and test sets, and five-fold cross validation. Assessed by five-fold validation, the random forest regression (RF) and radial basis function (RBF) models demonstrated better prediction performance in eight attempted machine learning models. The CLtot values predicted by the RF and RBF models were within two-fold of the obsd. values for 67.7 and 71.9% of cluster-split test set compds., resp., while the predictivity was worse in the time-split dataset. The predictivity of both models tended to be improved by incorporating in vitro parameters, unbound fraction in plasma (fu,p), and CLint. CLtot prediction utilizing in vitro CLint and the well-stirred model, correcting for the fraction unbound in blood, was substantially worse compared to machine learning approaches for the same cluster-split test set. The reason that CLtot is underestimated by IVIVE is not fully explained by considering the calcd. microsomal unbound fraction (cfu,mic), extended clearance classification system (ECCS), and omitting high clearance compds. in excess of hepatic blood flow. The anal. suggests that in silico machine learning models may have the power to reduce reliance on or replace in vitro and in vivo studies for chem. structure optimization in early drug discovery.(b) Miljković, F.; Martinsson, A.; Obrezanova, O.; Williamson, B.; Johnson, M.; Sykes, A.; Bender, A.; Greene, N. Machine Learning Models for Human In Vivo Pharmacokinetic Parameters with In-House Validation. Mol. Pharm. 2021, 18, 4520– 4530, DOI: 10.1021/acs.molpharmaceut.1c0071810bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXisVGrsLvK&md5=2ee0de42033d3f12164fe12ceff2a63dMachine Learning Models for Human In Vivo Pharmacokinetic Parameters with In-House ValidationMiljkovic, Filip; Martinsson, Anton; Obrezanova, Olga; Williamson, Beth; Johnson, Martin; Sykes, Andy; Bender, Andreas; Greene, NigelMolecular Pharmaceutics (2021), 18 (12), 4520-4530CODEN: MPOHBP; ISSN:1543-8384. (American Chemical Society)Prior to clin. development, a comprehensive pharmacokinetic characterization of a novel drug is required to understand its exposure at the site of action and elimination. Accordingly, in vitro assays and animal pharmacokinetic studies are regularly employed to predict drug exposure in humans, which is often costly and time-consuming. For this reason, the prediction of human pharmacokinetics at the point of design would be of high value for drug discovery. Therefore, we have established a comprehensive data curation protocol that enables machine learning evaluation of 12 human in vivo pharmacokinetic parameters using only chem. structure information and available doses for 1001 unique compds. These machine learning models were thoroughly investigated and validated using both an independent hold-out test set and AstraZeneca clin. data. In addn., the availability of preclin. predictions for a subset of internal clin. candidates allowed us to compare our in silico approach with state-of-the-art pharmacokinetic predictions. Based on this evaluation, three fit-for-purpose models for AUC PO (Rtest2 = 0.63; RMSEtest = 0.76), Cmax PO (Rtest2 = 0.68; RMSEtest = 0.62), and Vdss IV (Rtest2 = 0.47; RMSEtest = 0.50) were identified. Based on the findings, our machine learning models have considerable potential for practical applications in drug discovery, such as influencing decision-making in drug discovery projects and progression of drug candidates toward the clinic.
- 11Iwata, H.; Matsuo, T.; Mamada, H.; Motomura, T.; Matsushita, M.; Fujiwara, T.; Kazuya, M.; Handa, K. Prediction of total drug clearance in humans using animal data: proposal of a multimodal learning method based on deep learning. J. Pharm. Sci. 2021, 110, 1834, DOI: 10.1016/j.xphs.2021.01.02011https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXjs1yrt7w%253D&md5=6be4060a2333878006a36762ecdcd00bPrediction of Total Drug Clearance in Humans Using Animal Data: Proposal of a Multimodal Learning Method Based on Deep LearningIwata, Hiroaki; Matsuo, Tatsuru; Mamada, Hideaki; Motomura, Takahisa; Matsushita, Mayumi; Fujiwara, Takeshi; Kazuya, Maeda; Handa, KoichiJournal of Pharmaceutical Sciences (Philadelphia, PA, United States) (2021), 110 (4), 1834-1841CODEN: JPMSAE; ISSN:0022-3549. (Elsevier Inc.)Research into pharmacokinetics plays an important role in the development process of new drugs. Accurately predicting human pharmacokinetic parameters from preclin. data can increase the success rate of clin. trials. Since clearance (CL) which indicates the capacity of the entire body to process a drug is one of the most important parameters, many methods have been developed. However, there are still rooms to be improved for practical use in drug discovery research; improving CL prediction accuracy and understanding the chem. structure of compds. in terms of pharmacokinetics. To improve those, this research proposes a multimodal learning method based on deep learning that takes not only the chem. structure of a drug but also rat CL as inputs. Good results were obtained compared with the conventional animal scale-up method; the geometric mean fold error was 2.68 and the proportion of compds. with prediction errors of 2-fold or less was 48.5%. Furthermore, it was found to be possible to infer the partial structure useful for CL prediction by a structure contributing factor inference method. The validity of these results of structural interpretation of metabolic stability was confirmed by chemists.
- 12Troyanskaya, O.; Cantor, M.; Sherlock, G.; Brown, P.; Hastie, T.; Tibshirani, R.; Botstein, D.; Altman, R. B. Missing value estimation methods for DNA microarrays. Bioinformatics 2001, 17, 520– 525, DOI: 10.1093/bioinformatics/17.6.52012https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXltFOgsLY%253D&md5=d2c973e7f6355dec54e7fd2c9a599eb1Missing value estimation methods for DNA microarraysTroyanskaya, Olga; Cantor, Michael; Sherlock, Gavin; Brown, Pat; Hastie, Trevor; Tibshirani, Robert; Botstein, David; Altman, Russ B.Bioinformatics (2001), 17 (6), 520-525CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)Motivation: Gene expression microarray expts. can generate data sets with multiple missing expression values. Unfortunately, many algorithms for gene expression anal. require a complete matrix of gene array values as input. For example, methods such as hierarchical clustering and K-means clustering are not robust to missing data, and may lose effectiveness even with a few missing values. Methods for imputing missing data are needed, therefore, to minimize the effect of incomplete data sets on analyses, and to increase the range of data sets to which these algorithms can be applied. In this report, we investigate automated methods for estg. missing data. Results: We present a comparative study of several methods for the estn. of missing values in gene microarray data. We implemented and evaluated three methods: a Singular Value Decompn. (SVD) based method (SVDimpute), weighted K-nearest neighbors (KNNimpute), and row av. We evaluated the methods using a variety of parameter settings and over different real data sets, and assessed the robustness of the imputation methods to the amt. of missing data over the range of 1-20% missing values. We show that KNNimpute appears to provide a more robust and sensitive method for missing value estn. than SVDimpute, and both SVDimpute and KNNimpute surpass the commonly used row av. method (as well as filling missing values with zeros). We report results of the comparative expts. and provide recommendations and tools for accurate estn. of missing microarray data under a variety of conditions.
- 13Schafer, J. L.; Olsen, M. K. Multiple imputation for multivariate missing-data problems: A data analyst’s perspective. Multivariate Behavioral Res. 1998, 33, 545– 571, DOI: 10.1207/s15327906mbr3304_513https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BC28rpsVOrsw%253D%253D&md5=bdaa91e85de9e3ec3259976d4b62d907Multiple Imputation for Multivariate Missing-Data Problems: A Data Analyst's PerspectiveSchafer J L; Olsen M KMultivariate behavioral research (1998), 33 (4), 545-71 ISSN:0027-3171.Analyses of multivariate data are frequently hampered by missing values. Until recently, the only missing-data methods available to most data analysts have been relatively ad1 hoc practices such as listwise deletion. Recent dramatic advances in theoretical and computational statistics, however, have produced anew generation of flexible procedures with a sound statistical basis. These procedures involve multiple imputation (Rubin, 1987), a simulation technique that replaces each missing datum with a set of m > 1 plausible values. The rn versions of the complete data are analyzed by standard complete-data methods, and the results are combined using simple rules to yield estimates, standard errors, and p-values that formally incorporate missing-data uncertainty. New computational algorithms and software described in a recent book (Schafer, 1997a) allow us to create proper multiple imputations in complex multivariate settings. This article reviews the key ideas of multiple imputation, discusses the software programs currently available, and demonstrates their use on data from the Adolescent Alcohol Prevention Trial (Hansen & Graham, 199 I).
- 14Stekhoven, D. J.; Buhlmann, P. MissForest--non-parametric missing value imputation for mixed-type data. Bioinformatics 2012, 28, 112– 118, DOI: 10.1093/bioinformatics/btr59714https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs1yms7fO&md5=44a690989dad7424d50a1662f5de2625MissForest-non-parametric missing value imputation for mixed-type dataStekhoven, Daniel J.; Buehlmann, PeterBioinformatics (2012), 28 (1), 112-118CODEN: BOINFP; ISSN:1367-4803. (Oxford University Press)Modern data acquisition based on high-throughput technol. is often facing the problem of missing data. Algorithms commonly used in the anal. of such large-scale data often depend on a complete set. Missing value imputation offers a soln. to this problem. However, the majority of available imputation methods are restricted to one type of variable only: continuous or categorical. For mixed-type data, the different types are usually handled sep. Therefore, these methods ignore possible relations between variable types. We propose a non-parametric method which can cope with different types of variables simultaneously. Results: We compare several state of the art methods for the imputation of missing values. We propose and evaluate an iterative imputation method (missForest) based on a random forest. By averaging over many unpruned classification or regression trees, random forest intrinsically constitutes a multiple imputation scheme. Using the built-in out-of-bag error ests. of random forest, we are able to est. the imputation error without the need of a test set. Evaluation is performed on multiple datasets coming from a diverse selection of biol. fields with artificially introduced missing values ranging from 10% to 30%. We show that missForest can successfully handle missing values, particularly in datasets including different types of variables. In our comparative study, missForest outperforms other methods of imputation esp. in data settings where complex interactions and non-linear relations are suspected. The out-of-bag imputation error ests. of missForest prove to be adequate in all settings. Addnl., missForest exhibits attractive computational efficiency and can cope with high-dimensional data.
- 15Sawada, R.; Iwata, H.; Mizutani, S.; Yamanishi, Y. Target-Based Drug Repositioning Using Large-Scale Chemical-Protein Interactome Data. J. Chem. Inf. Model. 2015, 55, 2717– 2730, DOI: 10.1021/acs.jcim.5b0033015https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2MXhvVGhs73M&md5=5a1ee95bc842228b878d644d692fd798Target-Based Drug Repositioning Using Large-Scale Chemical-Protein Interactome DataSawada, Ryusuke; Iwata, Hiroaki; Mizutani, Sayaka; Yamanishi, YoshihiroJournal of Chemical Information and Modeling (2015), 55 (12), 2717-2730CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Drug repositioning, or the identification of new indications for known drugs, is a useful strategy for drug discovery. In this study, the authors developed novel computational methods to predict potential drug targets and new drug indications for systematic drug repositioning using large-scale chem.-protein interactome data. The authors explored the target space of drugs (including primary targets and off-targets) based on chem. structure similarity and phenotypic effect similarity by making optimal use of millions of compd.-protein interactions. On the basis of the target profiles of drugs, the authors constructed statistical models to predict new drug indications for a wide range of diseases with various mol. features. The proposed method outperformed previous methods in terms of interpretability, applicability, and accuracy. Finally, the authors conducted a comprehensive prediction of the drug-target-disease assocn. network for 8270 drugs and 1401 diseases and showed biol. meaningful examples of newly predicted drug targets and drug indications. The predictive model is useful to understand the mechanisms of the predicted drug indications.
- 16Martin, E. J.; Polyakov, V. R.; Zhu, X. W.; Tian, L.; Mukherjee, P.; Liu, X. All-Assay-Max2 pQSAR: Activity Predictions as Accurate as Four-Concentration IC50s for 8558 Novartis Assays. J. Chem. Inf. Model. 2019, 59, 4450– 4459, DOI: 10.1021/acs.jcim.9b0037516https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhslOqurbK&md5=00bbbad59334754948074d6e1e5edccbAll-Assay-Max2 pQSAR: Activity Predictions as Accurate as Four-Concentration IC50s for 8558 Novartis AssaysMartin, Eric J.; Polyakov, Valery R.; Zhu, Xiang-Wei; Tian, Li; Mukherjee, Prasenjit; Liu, XinJournal of Chemical Information and Modeling (2019), 59 (10), 4450-4459CODEN: JCISD8; ISSN:1549-9596. (American Chemical Society)Profile-quant. structure-activity relationship (pQSAR) is a massively multitask, two-step machine learning method with unprecedented scope, accuracy, and applicability domain. In step one, a "profile" of conventional single-assay random forest regression models are trained on a very large no. of biochem. and cellular pIC50 assays using Morgan 2 substructural fingerprints as compd. descriptors. In step two, a panel of partial least squares (PLS) models are built using the profile of pIC50 predictions from those random forest regression models as compd. descriptors (hence the name). Previously described for a panel of 728 biochem. and cellular kinase assays, we have now built an enormous pQSAR from 11 805 diverse Novartis (NVS) IC50 and EC50 assays. This large no. of assays, and hence of compd. descriptors for PLS, dictated reducing the profile by only including random forest regression models whose predictions correlate with the assay being modeled. The random forest regression and pQSAR models were evaluated with our "realistically novel" held-out test set, whose median av. similarity to the nearest training set member across the 11 805 assays was only 0.34, comparable to the novelty of compds. actually selected from virtual screens. For the 11 805 single-assay random forest regression models, the median correlation of prediction with the expt. was only rext2 = 0.05, virtually random, and only 8% of the models achieved our std. success threshold of rext2 = 0.30. For pQSAR, the median correlation was rext2 = 0.53, comparable to four-concn. exptl. IC50s, and 72% of the models met our rext2 > 0.30 std., totaling 8558 successful models. The successful models included assays from all of the 51 annotated target subclasses, as well as 4196 phenotypic assays, indicating that pQSAR can be applied to virtually any disease area. Every month, all models are updated to include new measurements, and predictions are made for 5.5 million NVS compds., totaling 50 billion predictions. Common uses have included virtual screening, selectivity design, toxicity and promiscuity prediction, mechanism-of-action prediction, and others. Several such actual applications are described.
- 17Maruhashi, K.; Todoriki, M.; Ohwa, T.; Goto, K.; Hasegawa, Y.; Inakoshi, H.; Anai, H. Learning Multi-way Relations Via Tensor Decomposition with Neural Networks, In Thirty-Second AAAI Conference on Artificial Intelligence , 2018.There is no corresponding record for this reference.
- 18Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 2012, 40, D1100– D1107, DOI: 10.1093/nar/gkr77718https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhs12htbjN&md5=aedf7793e1ca54b6a4fa272ea3ef7d0eChEMBL: a large-scale bioactivity database for drug discoveryGaulton, 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.
- 19Kim, S.; Thiessen, P. A.; Bolton, E. E.; Chen, J.; Fu, G.; Gindulyte, A.; Han, L.; He, J.; He, S.; Shoemaker, B. A.; Wang, J.; Yu, B.; Zhang, J.; Bryant, S. H. PubChem Substance and Compound databases. Nucleic Acids Res. 2016, 44, D1202– 1213, DOI: 10.1093/nar/gkv95119https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXhtV2gu7bE&md5=1ba53f15667506b761d05f0f02313892PubChem substance and compound databasesKim, Sunghwan; Thiessen, Paul A.; Bolton, Evan E.; Chen, Jie; Fu, Gang; Gindulyte, Asta; Han, Lianyi; He, Jane; He, Siqian; Shoemaker, Benjamin A.; Wang, Jiyao; Yu, Bo; Zhang, Jian; Bryant, Stephen H.Nucleic Acids Research (2016), 44 (D1), D1202-D1213CODEN: NARHAD; ISSN:0305-1048. (Oxford University Press)PubChem (https://pubchem.ncbi.nlm.nih.gov) is a public repository for information on chem. substances and their biol. activities, launched in 2004 as a component of the Mol. Libraries Roadmap Initiatives of the US National Institutes of Health (NIH). For the past 11 years, PubChem has grown to a sizable system, serving as a chem. information resource for the scientific research community. PubChem consists of three inter-linked databases, Substance, Compd. and BioAssay. The Substance database contains chem. information deposited by individual data contributors to PubChem, and the Compd. database stores unique chem. structures extd. from the Substance database. Biol. activity data of chem. substances tested in assay expts. are contained in the BioAssay database. This paper provides an overview of the PubChem Substance and Compd. databases, including data sources and contents, data organization, data submission using PubChem Upload, chem. structure standardization, web-based interfaces for textual and non-textual searches, and programmatic access. It also gives a brief description of PubChem3D, a resource derived from theor. three-dimensional structures of compds. in PubChem, as well as PubChemRDF, Resource Description Framework (RDF)-formatted PubChem data for data sharing, anal. and integration with information contained in other databases.
- 20Wishart, D. S.; Feunang, Y. D.; Guo, A. C.; Lo, E. J.; Marcu, A.; Grant, J. R.; Sajed, T.; Johnson, D.; Li, C.; Sayeeda, Z.; Assempour, N.; Iynkkaran, I.; Liu, Y.; Maciejewski, A.; Gale, N.; Wilson, A.; Chin, L.; Cummings, R.; Le, D.; Pon, A.; Knox, C.; Wilson, M. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 2018, 46, D1074– D1082, DOI: 10.1093/nar/gkx103720https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1cXitlGisbvI&md5=986b28c7ea546596a26dd3ba38f05feeDrugBank 5.0: a major update to the DrugBank database for 2018Wishart, David S.; Feunang, Yannick D.; Guo, An C.; Lo, Elvis J.; Marcu, Ana; Grant, Jason R.; Sajed, Tanvir; Johnson, Daniel; Li, Carin; Sayeeda, Zinat; Assempour, Nazanin; Iynkkaran, Ithayavani; Liu, Yifeng; Maciejewski, Adam; Gale, Nicola; Wilson, Alex; Chin, Lucy; Cummings, Ryan; Le, Diana; Pon, Allison; Knox, Craig; Wilson, MichaelNucleic Acids Research (2018), 46 (D1), D1074-D1082CODEN: NARHAD; ISSN:1362-4962. (Oxford University Press)DrugBank is a web-enabled database contg. comprehensivemol. information about drugs, their mechanisms, their interactions and their targets. First described in 2006, Drug- Bank has continued to evolve over the past 12 years in response to marked improvements to web stds. and changing needs for drug research and development. This year's update, DrugBank 5.0, represents the most significant upgrade to the database in more than 10 years. In many cases, existing data content has grown by 100% or more over the last update. For instance, the total no. of investigational drugs in the database has grown by almost 300%, the no. of drug-drug interactions has grown by nearly 600% and the no. of SNP-assocd. drug effects has grown more than 3000%. Significant improvements have been made to the quantity, quality and consistency of drug indications, drug binding data as well as drug-drug and drug-food interactions. A great deal of brand new data have also been added to DrugBank 5.0. This includes information on the influence of hundreds of drugs on metabolite levels (pharmacometabolomics), gene expression levels (pharmacotranscriptomics) and protein expression levels (pharmacoproteomics). New data have also been added on the status of hundreds of newdrug clin. trials and existing drug repurposing trials. Many other important improvements in the content, interface and performance of the DrugBank website have been made and these should greatly enhance its ease of use, utility and potential applications in many areas of pharmacol. research, pharmaceutical science and drug education.
- 21Varma, M. V. S.; Feng, B.; Obach, R. S.; Troutman, M. D.; Chupka, J.; Miller, H. R.; El-Kattan, A. Physicochemical determinants of human renal clearance. J. Med. Chem. 2009, 52, 4844– 4852, DOI: 10.1021/jm900403j21https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1MXlvFOnt7s%253D&md5=a99ccaa1651d04ed7f7c0a49679f55f6Physicochemical Determinants of Human Renal ClearanceVarma, Manthena V. S.; Feng, Bo; Obach, R. Scott; Troutman, Matthew D.; Chupka, Jonathan; Miller, Howard R.; El-Kattan, AymanJournal of Medicinal Chemistry (2009), 52 (15), 4844-4852CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)Kidney plays an important role in the elimination of drugs, esp. with low or negligible hepatic clearance. An anal. of the interrelation of physicochem. properties and the human renal clearance for a data set of 391 drugs or compds. tested in humans is presented. The data set indicated that lipophilicity shows a neg. relationship while polar descriptors show a pos. relationship with renal clearance. Anal. of net secreted and net reabsorbed subsets revealed that hydrophilic ionized compds. are probable compds. to show net secretion and a possible drug-drug interaction due to their likely interaction with uptake transporters and inherent low passive reabsorption. The physicochem. space and renal clearance were also statistically analyzed by therapeutic area. In conclusion, ionization state, lipophilicity, and polar descriptors are found to be the physicochem. determinants of renal clearance. These fundamental properties can be valuable in early prediction of human renal clearance and can aid the chemist in structural modifications to optimize drug disposition.
- 22Jahandideh-Tehrani, M.; Bozorg-Haddad, O.; Loaiciga, H. A. Application of particle swarm optimization to water management: an introduction and overview. Environ. Monit. Assess. 2020, 192, 281, DOI: 10.1007/s10661-020-8228-z22https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB38zkslWqsg%253D%253D&md5=9b05b68e5c836b1fffd6cd2b0069d6f7Application of particle swarm optimization to water management: an introduction and overviewJahandideh-Tehrani Mahsa; Bozorg-Haddad Omid; Loaiciga Hugo AEnvironmental monitoring and assessment (2020), 192 (5), 281 ISSN:.Particle swarm optimization (PSO) is a stochastic population-based optimization algorithm inspired by the interactions of individuals in a social world. This algorithm is widely applied in different fields of water resources problems. This paper presents a comprehensive overview of the basic PSO algorithm search strategy and PSO's applications and performance analysis in water resources engineering optimization problems. Our literature review revealed 22 different varieties of the PSO algorithm. The characteristics of each PSO variety together with their applications in different fields of water resources engineering (e.g., reservoir operation, rainfall-runoff modeling, water quality modeling, and groundwater modeling) are highlighted. The performances of different PSO variants were compared with other evolutionary algorithms (EAs) and mathematical optimization methods. The review evaluates the capability and comparative performance of PSO variants over conventional EAs (e.g., simulated annealing, differential evolution, genetic algorithm, and shark algorithm) and mathematical methods (e.g., support vector machine and differential dynamic programming) in terms of proper convergence to optimal Pareto fronts, faster convergence rate, and diversity of computed solutions.
- 23(a) Breiman, L. Random forests. Mach. Learn. 2001, 45, 5– 32, DOI: 10.1023/A:1010933404324There is no corresponding record for this reference.(b) Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V. Scikit-learn: Machine learning in Python. J. Machine Learning Res. 2011, 12, 2825– 2830There is no corresponding record for this reference.
- 24Glorot, X.; Bordes, A.; Bengio, Y. In Deep Sparse Rectifier Neural Networks . Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics2011; pp 315– 323.There is no corresponding record for this reference.
- 25Ioffe, S.; Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift . arXiv preprint arXiv:1502.03167 2015.There is no corresponding record for this reference.
- 26Srivastava, N.; Hinton, G.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overfitting. J. Machine Learning Res. 2014, 15, 1929– 1958There is no corresponding record for this reference.
- 27Varoquaux, G.; Buitinck, L.; Louppe, G.; Grisel, O.; Pedregosa, F.; Mueller, A. Scikit-learn: Machine learning without learning the machinery. GetMobile: Mobile Comput. Commun. 2015, 19, 29– 33, DOI: 10.1145/2786984.2786995There is no corresponding record for this reference.
- 28Tang, H.; Mayersohn, M. A novel model for prediction of human drug clearance by allometric scaling. Drug Metab. Dispos. 2005, 33, 1297– 1303, DOI: 10.1124/dmd.105.00414328https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2MXpsFamtrc%253D&md5=4592ad4e0aba395d3755bdefc50209f5A novel model for prediction of human drug clearance by allometric scalingTang, Huadong; Mayersohn, MichaelDrug Metabolism and Disposition (2005), 33 (9), 1297-1303CODEN: DMDSAI; ISSN:0090-9556. (American Society for Pharmacology and Experimental Therapeutics)Sixty-one sets of clearance (CL) values in animal species were allometrically scaled for predicting human clearance. Unbound fractions (fu) of drug in plasma in rats and humans were obtained from the literature. A model was developed to predict human CL: CL = 33.35 mL/min × (a/Rfu)0.770, where Rfu is the fu ratio between rats and humans and a is the coeff. obtained from allometric scaling. The new model was compared with simple allometric scaling and the "rule of exponents" (ROE). Results indicated that the new model provided better predictability for human values of CL than did ROE. It is esp. significant that for the first time the proposed model improves the prediction of CL for drugs illustrating large vertical allometry.
- 29(a) Jones, R. D.; Jones, H. M.; Rowland, M.; Gibson, C. R.; Yates, J. W.; Chien, J. Y.; Ring, B. J.; Adkison, K. K.; Ku, M. S.; He, H. PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 2: comparative assessment of prediction methods of human volume of distribution. J. Pharm. Sci. 2011, 100, 4074– 4089, DOI: 10.1002/jps.2255329ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3MXhtVylur%252FE&md5=7c3bf546e194856406b6ee569430c48bPhRMA CPCDC initiative on predictive models of human pharmacokinetics, Part 2: Comparative assessment of prediction methods of human volume of distributionJones, Rhys Do; Jones, Hannah M.; Rowland, Malcolm; Gibson, Christopher R.; Yates, James W. T.; Chien, Jenny Y.; Ring, Barbara J.; Adkison, Kimberly K.; Ku, M. Sherry; He, Handan; Vuppugalla, Ragini; Marathe, Punit; Fischer, Volker; Dutta, Sandeep; Sinha, Vikash K.; Bjoernsson, Thorir; Lave, Thierry; Poulin, PatrickJournal of Pharmaceutical Sciences (2011), 100 (10), 4074-4089CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)The objective of this study was to evaluate the performance of various empirical, semimechanistic and mechanistic methodologies with and without protein binding corrections for the prediction of human vol. of distribution at steady state (Vss). PhRMA member companies contributed a set of blinded data from preclin. and clin. studies, and 18 drugs with i.v. clin. pharmacokinetics (PK) data were available for the anal. In vivo and in vitro preclin. data were used to predict Vss by 24 different methods. Various statistical and outlier techniques were employed to assess the predictability of each method. There was not simply one method that predicts Vss accurately for all compds. Across methods, the max. success rate in predicting human Vss was 100%, 94%, and 78% of the compds. with predictions falling within tenfold, threefold, and twofold error, resp., of the obsd. Vss. Generally, the methods that made use of in vivo preclin. data were more predictive than those methods that relied solely on in vitro data. However, for many compds., in vivo data from only two species (generally rat and dog) were available and/or the required in vitro data were missing, which meant some methods could not be properly evaluated. It is recommended to initially use the in vitro tissue compn.-based equations to predict Vss in preclin. species and humans, putting the assumptions and compd. properties into context. As in vivo data become available, these predictions should be reassessed and rationalized to indicate the level of confidence (uncertainty) in the human Vss prediction. The top three methods that perform strongly at integrating in vivo data in this way were the Oie-Tozer, the rat -dog-human proportionality equation, and the lumped-PBPK approach. Overall, the scientific benefit of this study was to obtain greater characterization of predictions of human Vss from several methods available in the literature. © 2011 Wiley-Liss, Inc. and the American Pharmacists Assocn. J Pharm Sci 100:4074-4089, 2011.(b) Obach, R. S.; Baxter, J. G.; Liston, T. E.; Silber, B. M.; Jones, B. C.; Macintyre, F.; Rance, D. J.; Wastall, P. The prediction of human pharmacokinetic parameters from preclinical and in vitro metabolism data. J. Pharmacol. Exp. Ther. 1997, 283, 46– 5829bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2sXmslCltr0%253D&md5=68d1910d925e26ec4a8ef23043ffb1edThe prediction of human pharmacokinetic parameters from preclinical and in vitro metabolism dataObach, R. Scott; Baxter, James G.; Liston, Theodore E.; Silber, B. Michael; Jones, Barry C.; Macintyre, Flona; Rance, David J.; Wastall, PhilipJournal of Pharmacology and Experimental Therapeutics (1997), 283 (1), 46-58CODEN: JPETAB; ISSN:0022-3565. (Williams & Wilkins)A review with 35 refs. We describe a comprehensive retrospective anal. in which the abilities of several methods by which human pharmacokinetic parameters are predicted from preclin. pharmacokinetic data and/or in vitro metab. data were assessed. The prediction methods examd. included both methods from the scientific literature as well as some described in this report for the first time. Four methods were examd. for their ability to predict human vol. of distribution. Three were highly predictive, yielding, on av., predictions that were within 60% to 90% of actual values. Twelve methods were assessed for their utility in predicting clearance. The most successful allometric scaling method yielded clearance predictions that were, on av., within 80% of actual values. The best methods in which in vitro metab. data from human liver microsomes were scaled to in vivo clearance values yielded predicted clearance values that were, on av., within 70% to 80% of actual values. Human t1/2 was predicted by combining predictions of human vol. of distribution and clearance. The best t1/2 prediction methods successfully assigned compds. to appropriate dosing regimen categories (e.g., once daily, twice daily and so forth) 70% to 80% of the time. In addn., correlations between human t1/2 and t1/2 values from preclin. species were also generally successful (72-87%) when used to predict human dosing regimens. In summary, this retrospective anal. has identified several approaches by which human pharmacokinetic data can be predicted from preclin. data. Such approaches should find utility in the drug discovery and development processes in the identification and selection of compds. that will possess appropriate pharmacokinetic characteristics in humans for progression to clin. trials.
- 30Słoczyńska, K.; Gunia-Krzyzak, A.; Koczurkiewicz, P.; Wojcik-Pszczola, K.; Zelaszczyk, D.; Popiol, J.; Pekala, E. Metabolic stability and its role in the discovery of new chemical entities. Acta. Pharm. 2019, 69, 345– 361, DOI: 10.2478/acph-2019-002430https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXntFWhtg%253D%253D&md5=2dfad07056bce4e4cc32264c8ce07040Metabolic stability and its role in the discovery of new chemical entitiesSloczynska, Karolina; Gunia-Krzyzak, Agnieszka; Koczurkiewicz, Paulina; Wojcik-Pszczola, Katarzyna; Zelaszczyk, Dorota; Popiol, Justyna; Pekala, ElzbietaActa Pharmaceutica (Warsaw, Poland) (2019), 69 (3), 345-361CODEN: ACPHEE; ISSN:1846-9558. (Sciendo)A review. Detn. of metabolic profiles of new chem. entities is a key step in the process of drug discovery, since it influences pharmacokinetic characteristics of therapeutic compds. One of the main challenges of medicinal chem. is not only to design compds. demonstrating beneficial activity, but also mols. exhibiting favorable pharmacokinetic parameters. Chem. compds. can be divided into those which are metabolized relatively fast and those which undergo slow biotransformation. Rapid biotransformation reduces exposure to the maternal compd. and may lead to the generation of active, non-active or toxic metabolites. In contrast, high metabolic stability may promote interactions between drugs and lead to parent compd. toxicity. In the present paper, issues of compd. metabolic stability will be discussed, with special emphasis on its significance, in vitro metabolic stability testing, dilemmas regarding in vitro-in vivo extrapolation of the results and some aspects relating to different preclin. species used in in vitro metabolic stability assessment of compds.
- 31Wakayama, N.; Toshimoto, K.; Maeda, K.; Hotta, S.; Ishida, T.; Akiyama, Y.; Sugiyama, Y. In Silico Prediction of Major Clearance Pathways of Drugs among 9 Routes with Two-Step Support Vector Machines. Pharm. Res. 2018, 35, 197, DOI: 10.1007/s11095-018-2479-131https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A280%3ADC%252BB3c3hs12mtw%253D%253D&md5=94b5e6b45c2eb894cdcc4abe0d7b0f81In Silico Prediction of Major Clearance Pathways of Drugs among 9 Routes with Two-Step Support Vector MachinesWakayama Naomi; Toshimoto Kota; Hotta Shun; Ishida Takashi; Akiyama Yutaka; Toshimoto Kota; Sugiyama Yuichi; Maeda KazuyaPharmaceutical research (2018), 35 (10), 197 ISSN:.PURPOSE: The clearance pathways of drugs are critical elements for understanding the pharmacokinetics of drugs. We previously developed in silico systems to predict the five clearance pathway using a rectangular method and a support vector machine (SVM). In this study, we improved our classification system by increasing the number of clearance pathways available for our prediction (CYP1A2, CYP2C8, CYP2C19, and UDP-glucuronosyl transferases (UGTs)) and by accepting multiple major pathways. METHODS: Using the four default descriptors (charge, molecular weight, logD at pH 7.0, and unbound fraction in plasma), three kinds of SVM-based predictors based on traditional single-step approach or two-step focusing approaches with subset or partition clustering were developed. The two-step approach with subset clustering resulted in the highest prediction performance. The feature-selection of additional descriptors based on a greedy algorithm was employed to further improve the predictability. RESULTS: The prediction accuracy for each pathway was increased to more than 0.83 with the exception of CYP2C19 and UGTs pathways, whose accuracies were below 0.7. Prediction performance of CYP1A2, CYP3A4 and renal excretion pathways were found to be acceptable using external dataset. CONCLUSIONS: We successfully constructed a novel SVM-based predictor for the multiple major clearance pathways based on chemical structures.
- 32Poulin, P.; Dambach, D. M.; Hartley, D. H.; Ford, K.; Theil, F. P.; Harstad, E.; Halladay, J.; Choo, E.; Boggs, J.; Liederer, B. M.; Dean, B.; Diaz, D. An algorithm for evaluating potential tissue drug distribution in toxicology studies from readily available pharmacokinetic parameters. J. Pharm. Sci. 2013, 102, 3816– 3829, DOI: 10.1002/jps.2367032https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC3sXhtFCitLvM&md5=61c57bce0b889c7dab87369a8b186b0eAn Algorithm for Evaluating PotentKpial Tissue Drug Distribution in Toxicology Studies from Readily Available Pharmacokinetic ParametersPoulin, Patrick; Dambach, Donna M.; Hartley, Dylan H.; Ford, Kevin; Theil, Frank-Peter; Harstad, Eric; Halladay, Jason; Choo, Edna; Boggs, Jason; Liederer, Bianca M.; Dean, Brian; Diaz, DoloresJournal of Pharmaceutical Sciences (2013), 102 (10), 3816-3829CODEN: JPMSAE; ISSN:0022-3549. (John Wiley & Sons, Inc.)Having an understanding of drug tissue accumulation can be informative in the assessment of target organ toxicities; however, obtaining tissue drug levels from toxicol. studies by bioanal. methods is labor-intensive and infrequently performed. Addnl., there are no described methods for predicting tissue drug distribution for the exptl. conditions in toxicol. studies, which typically include non-steady-state conditions and very high exposures that may sat. several processes. The aim was the development of an algorithm to provide semiquant. and quant. ests. of tissue-to-plasma concn. ratios (Kp) for several tissues from readily available parameters of pharmacokinetics (PK) such as vol. of distribution (Vd) and clearance of each drug, without performing tissue measurement in vivo. The computational approach is specific for the oral route of administration and non-steady-state conditions and was applied for a dataset of 29 Genentech small mols. such as neutral compds. as well as weak and strong org. bases. The max. success rate in predicting Kp values within 2.5-fold error of obsd. Kp values was 82% at low doses (<100 mg/kg) in preclin. species. Prediction accuracy was relatively lower with satn. at high doses (≥100 mg/kg); however, an approach to perform low-to-high dose extrapolations of Kp values was presented and applied successfully in most cases. An approach for the interspecies scaling was also applied successfully. Finally, the proposed algorithm was used in a case study and successfully predicted differential tissue distribution of two small-mol. MET kinase inhibitors, which had different toxicity profiles in mice. This newly developed algorithm can be used to predict the partition coeffs. Kp for small mols. in toxicol. studies, which can be leveraged to optimize the PK drivers of tissue distribution in an attempt to decrease drug tissue level, and improve safety margins. © 2013 Wiley Periodicals, Inc. and the American Pharmacists Assocn. J Pharm Sci.
- 33(a) Poulin, P.; Theil, F. P. A priori prediction of tissue: plasma partition coefficients of drugs to facilitate the use of physiologically-based pharmacokinetic models in drug discovery. J. Pharm. Sci. 2000, 89, 16– 35, DOI: 10.1002/(SICI)1520-6017(200001)89:1<16::AID-JPS3>3.0.CO;2-E33ahttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3cXit1eis7c%253D&md5=c72666192591b0768bec8aaed1608cf8A priori prediction of tissue:plasma partition coefficients of drugs to facilitate the use of physiologically-based pharmacokinetic models in drug discoveryPoulin, Patrick; Theil, Frank-PeterJournal of Pharmaceutical Sciences (2000), 89 (1), 16-35CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)The tissue:plasma (Pt:p) partition coeffs. (PCs) are important drug-specific input parameters in physiol. based pharmacokinetic (PBPK) models used to est. the disposition of drugs in biota. Until now the use of PBPK models in early stages of the drug discovery process was not possible, since the estn. of Pt:p of new drug candidates by using conventional in vitro and/or in vivo methods is too time and cost intensive. The objectives of the study were (i) to develop and validate two mechanistic equations for predicting a priori the rabbit, rat and mouse Pt:p of non-adipose and non-excretory tissues (bone, brain, heart, intestine, lung, muscle, skin, spleen) for 65 structurally unrelated drugs and (ii) to evaluate the adequacy of using Pt:p of muscle as predictors for Pt:p of other tissues. The first equation predicts Pt:p at steady state, assuming a homogenous distribution and passive diffusion of drugs in tissues, from a ratio of soly. and macromol. binding between tissues and plasma. The ratio of soly. was estd. from log vegetable oil:water PCs (Kvo:w) of drugs and lipid and water levels in tissues and plasma, whereas the ratio of macromol. binding for drugs was estd. from tissue interstitial fluid-to-plasma concn. ratios of albumin, globulins and lipoproteins. The second equation predicts Pt:p of drugs residing predominantly in the interstitial space of tissues. Therefore, the fractional vol. content of interstitial space in each tissue replaced drug solubilities in the first equation. Following the development of these equations, regression analyses between Pt:p of muscle and those of the other tissues were examd. The av. ratio of predicted-to-exptl. Pt:p values was 1.26 (SD = 1.40, r = 0.90), and 85% of the 269 predicted values were within a factor of three of the corresponding literature values obtained under in vivo and in vitro conditions. For predicted and exptl. Pt:p, linear relationships (r > 0.9 in most cases) were obsd. between muscle and other tissues, suggesting that Pt:p of muscle is a good predictor for the Pt:p of other tissues. The two previous equations could explain the mechanistic basis of these linear relationships. The practical aim of this study is a worthwhile goal for pharmacokinetic screening of new drug candidates.(b) Poulin, P.; Schoenlein, K.; Theil, F. P. Prediction of adipose tissue: plasma partition coefficients for structurally unrelated drugs. J. Pharm. Sci. 2001, 90, 436– 447, DOI: 10.1002/1520-6017(200104)90:4<436::AID-JPS1002>3.0.CO;2-P33bhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD3MXivVant7s%253D&md5=c9af9c58e95ca183fdd30f94ea367edbPrediction of adipose tissue:plasma partition coefficients for structurally unrelated drugsPoulin, Patrick; Schoenlein, Kerstin; Theil, Frank-PeterJournal of Pharmaceutical Sciences (2001), 90 (4), 436-447CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)Tissue:plasma (Pt:p) partition coeffs. (PCs) are important parameters describing tissue distribution of drugs. The ultimate goal in early drug discovery is to develop and validate in silico methods for predicting a priori the Pt:p for each new drug candidate. In this context, tissue compn.-based equations have recently been developed and validated for predicting a priori the non-adipose and adipose Pt:p for neutral org. solvents and pollutants. For ionizable drugs that bind to different degrees to common plasma proteins, only their non-adipose Pt:p values have been predicted with these equations. The only compd.-dependent input parameters for these equations are the lipophilicity parameter, such as olive oil-water PC (Kvo:w) or n-octanol-water PC (Po:w), and/or unbound fraction in plasma (fup) detd. under in vitro conditions. Tissue compn.-based equations could potentially also be used to predict adipose tissue-plasma PCs (Pat:p) for ionized drugs. The main objective of the present study was to modify these equations for predicting in vivo Pat:p (white fat) for 14 structurally unrelated ionized drugs that bind substantially to plasma macromols. in rats, rabbits, or humans. The second objective was to verify whether Kvo:w or Po:w provides more accurate predictions of in vivo Pat:p (i.e., to verify whether olive oil or n-octanol is the better surrogate for lipids in adipose tissue). The second objective was supported by comparing in vitro data on Pat:p with those on olive oil-plasma PC (Kvo:p) for five drugs. Furthermore, in vivo Pat:p was not only predicted from Kvo:w and Po:w of the non-ionized species, but also from Kvo:w* and Po:w*, taking into account the ionized species in addn. The Pat:p predicted from Kvo:w*, Po:w*, and Po:w differ from the in vivo Pat:p by an av. factor of 1.17 (SD = 0.44, r = 0.95), 15.0 (SD = 15.7, r = 0.59), and 40.7 (SD = 57.2, r = 0.33), resp. The in vitro values of Kvo:p differ from those of Pat:p by an av. factor of 0.86 (SD = 0.16, r = 0.99, n = 5). The results demonstrate that (i) the equation using only data on fup as input and olive oil as lipophilicity surrogate is able to provide accurate predictions of in vivo Pat:p, and (ii) olive oil is a better surrogate of the adipose tissue lipids than n-octanol. The present study is an innovative method for predicting in vivo fat partitioning of drugs in mammals.(c) Poulin, P.; Krishnan, K. A biologically-based algorithm for predicting human tissue: blood partition coefficients of organic chemicals. Hum. Exp. Toxicol. 1995, 14, 273– 280, DOI: 10.1177/09603271950140030733chttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2MXls1ersL0%253D&md5=131ccbfb41ef67bbf75ea2fcfd774e56A biologically-based algorithm for predicting human tissue: blood partition coefficients of organic chemicalsPoulin, Patrick; Krishnan, KannanHuman & Experimental Toxicology (1995), 14 (3), 273-80CODEN: HETOEA; ISSN:0960-3271.A biol.-based algorithm for predicting the tissue: blood partition coeffs. (PCs) of org. chems. has been developed. The approach consisted of (i) describing tissues and blood in terms of their neutral lipid, phospholipid, and water contents, (ii) obtaining data on the soly. of chems. in n-octanol and water, and (iii) calcg. the tissue: blood PCs by assuming that the soly. of a chem. in n-octanol corresponds to its soly. in neutral lipids, the soly. in water corresponds to the soly. in tissue/blood water fraction, and the soly. in phospholipids is a function of soly. in water and n-octanol. The adequacy of this approach was verified by comparing the predicted values with previously published exptl. data on human tissue (liver, lung, muscle, kidney, brain, adipose tissue): blood PCs for 23 org. chems. In the case of liver, lung, and muscle, the predicted PC values were in close agreement with the higher-end of the range of exptl. PC values found in the literature. The predicted brain: and kidney: blood PCs were greater than the exptl. PCs in most cases by approx. a factor of two. Whereas the adipose tissue: blood PCs of relatively less hydrophilic chems. were adequately predicted, the predicted PCs for relatively more hydrophilic chems. were much greater than the exptl.-detd. values. There was a good agreement between the predicted and exptl.-detd. blood soly. of the 23 chems. chosen for this study, indicating that the overestn. of tissue:blood PCs by the present method is not due to under-estn. of blood soly. of chems. Rather, it might be due to the lower tissue soly. of chems. obsd. exptl. due to the complexity of the tissue matrixes. This novel approach of describing tissues in terms of the type of lipid and water content should enable the prediction of the tissue:blood PCs of org. chems. with information on their soly. in water and n-octanol, for developing physiol.-based toxicokinetic models.(d) Berezhkovskiy, L. M. Volume of distribution at steady state for a linear pharmacokinetic system with peripheral elimination. J. Pharm. Sci. 2004, 93, 1628– 1640, DOI: 10.1002/jps.2007333dhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXks12hsrY%253D&md5=55af00dd6f68f348a617cbef9784757dVolume of distribution at steady state for a linear pharmacokinetic system with peripheral eliminationBerezhkovskiy, Leonid M.Journal of Pharmaceutical Sciences (2004), 93 (6), 1628-1640CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)A review. The problem of finding the steady-state vol. of distribution Vss for a linear pharmacokinetic system with peripheral drug elimination is considered. A commonly used equation Vss=(D/AUC)*MRT is applicable only for the systems with central (plasma) drug elimination. The following equation, Vss=(D/AUC)*MRTint, was obtained, where AUC is the commonly calcd. area under the time curve of the total drug concn. in plasma after i.v. administration of bolus drug dose, D, and MRTint is the intrinsic mean residence time, which is the av. time the drug spends in the body (system) after entering the systemic circulation (plasma). The value of MRTint cannot be found from a drug plasma concn. profile after an i.v. bolus drug input if a peripheral drug exit occurs. The obtained equation does not contain the assumption of an immediate equil. of protein and tissue binding in plasma and organs, and thus incorporates the rates of all possible reactions. If drug exits the system only through central compartment (plasma) and there is an instant equil. between bound and unbound drug fractions in plasma, then MRTint becomes equal to MRT=AUMC/AUC, which is calcd. using the time course of the total drug concn. in plasma after an i.v. bolus injection. Thus, the obtained equation coincides with the traditional one, Vss=(D/AUC)*MRT, if the assumptions for validity of this equation are met. Exptl. methods for detg. the steady-state vol. of distribution and MRTint, as well as the problem of detg. whether peripheral drug elimination occurs, are considered. The equation for calcn. of the tissue-plasma partition coeff. with the account of peripheral elimination is obtained. The difference between traditionally calcd. Vss=(D/AUC)*MRT and the true value given by (D/AUC)*MRTint is discussed.(e) Rodgers, T.; Rowland, M. Physiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterions. J. Pharm. Sci. 2006, 95, 1238– 1257, DOI: 10.1002/jps.2050233ehttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD28XlsFWlsrY%253D&md5=6691b676c3ac213a96c8724576ad761aPhysiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterionsRodgers, Trudy; Rowland, MalcolmJournal of Pharmaceutical Sciences (2006), 95 (6), 1238-1257CODEN: JPMSAE; ISSN:0022-3549. (Wiley-Liss, Inc.)A key component of whole-body physiol. based pharmacokinetic (WBPBPK) models is the tissue-to-plasma water partition coeffs. (Kpu's). The predictability of Kpu values using mechanistically derived equations has been investigated for 7 very weak bases, 20 acids, 4 neutral drugs and 8 zwitterions in rat adipose, bone, brain, gut, heart, kidney, liver, lung, muscle, pancreas, skin, spleen and thymus. These equations incorporate expressions for dissoln. in tissue water and for partitioning into neutral lipids and neutral phospholipids. Addnl., assocns. with acidic phospholipids were incorporated for zwitterions with a highly basic functionality or extracellular proteins for the other compd. classes. The affinity for these cellular constituents was detd. from blood cell data or plasma protein binding, resp. These equations assume drugs are passively distributed and that processes are nonsatg. Resultant Kpu predictions were more accurate when compared to published equations, with 84% as opposed to 61% of the predicted values agreeing with exptl. values to within a factor of 3. This improvement was largely due to the incorporation of distribution processes related to drug ionization, an issue that was not addressed in earlier equations. Such advancements in parameter prediction will assist WBPBPK modeling, where time, cost, and labor requirements greatly deter its application.(f) Schmitt, W. General approach for the calculation of tissue to plasma partition coefficients. Toxicol. In Vitro 2008, 22, 457– 467, DOI: 10.1016/j.tiv.2007.09.01033fhttps://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD1cXhsFygt7o%253D&md5=b2cdca7ad30692ad2ff57ffcbe86a775General approach for the calculation of tissue to plasma partition coefficientsSchmitt, WalterToxicology in Vitro (2008), 22 (2), 457-467CODEN: TIVIEQ; ISSN:0887-2333. (Elsevier Ltd.)A new mechanistic, universal model for the calcn. of steady state tissue:plasma partition coeffs. (Kt:p) of org. chems. in mammalian species was developed. The approach allows the estn. of Kt:p-values based on the compn. of the tissues in terms of water, neutral lipids, neutral and acidic phospholipids and proteins using the lipophilicity, the binding to phospholipid membranes, the pKa and the unbound fraction in blood plasma as compd. specific parameters. Taking explicitly into account the sign and fraction of the charge of the compds. at the physiol. pH the method is universally applicable to neutral, acidic, basic or multiply charged substances and has thus a significantly extended applicability compared to previously published approaches. The model was applied to 59 chem. diverse drug compds. for which tissue:plasma partition coeffs. are reported in the literature. In total 474 exptl. obsd. Kt:p values for 12 tissues and the red blood cells were available and could be compared to model results. For 73% of the calcd. values a deviation less than 3-fold from the resp. obsd. value was found, proving the validity of the approach.
- 34Tess, D. A.; Eng, H.; Kalgutkar, A. S.; Litchfield, J.; Edmonds, D. J.; Griffith, D. A.; Varma, M. V. S. Predicting the Human Hepatic Clearance of Acidic and Zwitterionic Drugs. J. Med. Chem. 2020, 63, 11831– 11844, DOI: 10.1021/acs.jmedchem.0c0103334https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhvFartbvN&md5=3e9b035730014972fa070f7cc4cd6404Predicting the Human Hepatic Clearance of Acidic and Zwitterionic DrugsTess, David A.; Eng, Heather; Kalgutkar, Amit S.; Litchfield, John; Edmonds, David J.; Griffith, David A.; Varma, Manthena V. S.Journal of Medicinal Chemistry (2020), 63 (20), 11831-11844CODEN: JMCMAR; ISSN:0022-2623. (American Chemical Society)Prospective predictions of human hepatic clearance for anionic/zwitterionic compds., which are oftentimes subjected to transporter-mediated uptake, are challenging in drug discovery. We evaluated the utility of preclin. species, rats and cynomolgus monkeys [nonhuman primates (NHPs)], to predict the human hepatic clearance using a diverse set of acidic/zwitterionic drugs. Preclin. clearance data were generated following i.v. dosing in rats/NHPs and compared to the human clearance data (n = 18/27). Single-species scaling of NHP clearance with an allometric exponent of 0.50 allowed for good prediction of human clearance (fold error ~ 2.1, bias ~ 1.0), with ~ 86% predictions within 3-fold. In comparison, rats underpredicted the clearance of lipophilic acids, while overprediction was noted for hydrophilic acids. Finally, an in vitro clearance assay based on human hepatocytes, which is routinely used in discovery setting, markedly underpredicted human clearance (bias ~ 0.12). Collectively, this study provides insights into the usefulness of the preclin. models in enabling pharmacokinetic optimization for acid/zwitterionic drug candidates.
- 35Miyamoto, M.; Iwasaki, S.; Chisaki, I.; Nakagawa, S.; Amano, N.; Hirabayashi, H. Comparison of predictability for human pharmacokinetics parameters among monkeys, rats, and chimeric mice with humanised liver. Xenobiotica 2017, 47, 1052– 1063, DOI: 10.1080/00498254.2016.126516035https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC2sXjs1Gksb0%253D&md5=442cba86ccdbe16bb08cee7ef572730bComparison of predictability for human pharmacokinetics parameters among monkeys, rats, and chimeric mice with humanised liverMiyamoto, Maki; Iwasaki, Shinji; Chisaki, Ikumi; Nakagawa, Sayaka; Amano, Nobuyuki; Hirabayashi, HidekiXenobiotica (2017), 47 (12), 1052-1063CODEN: XENOBH; ISSN:0049-8254. (Taylor & Francis Ltd.)The aim of the present study was to evaluate the usefulness of chimeric mice with humanised liver (PXB mice) for the prediction of clearance (CLt) and vol. of distribution at steady state (Vdss), in comparison with monkeys, which have been reported as a reliable model for human pharmacokinetics (PK) prediction, and with rats, as a conventional PK model. CLt and Vdss values in PXB mice, monkeys and rats were detd. following i.v. administration of 30 compds. known to be mainly eliminated in humans via the hepatic metab. by various drug-metabolizing enzymes. Using single-species allometric scaling, human CLt and Vdss values were predicted from the three animal models. Predicted CLt values from PXB mice exhibited the highest predictability: 25 for PXB mice, 21 for monkeys and 14 for rats were predicted within a three-fold range of actual values among 30 compds. For predicted human Vdss values, the no. of compds. falling within a three-fold range was 23 for PXB mice, 24 for monkeys, and 16 for rats among 29 compds. PXB mice indicated a higher predictability for CLt and Vdss values than the other animal models. These results demonstrate the utility of PXB mice in predicting human PK parameters.
- 36Russell, W.; Burch, R. The Principles of Humane Experimental Technique. Wheathampstead, Universities Federation for Animal Welfare: UK; 1959.There is no corresponding record for this reference.
Supporting Information
Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.2c00318.
The data used in this paper are listed in Supporting_information_dataset.xlsx (PDF)
Terms & Conditions
Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.