A Rapid Computational Filter for Cytochrome P450 1A2 Inhibition Potential of Compound Libraries

Chohan, Kamaldeep K.; Paine, Stuart W.; Mistry, Jaina; Barton, and Patrick; Davis, A. M.

doi:10.1021/jm048959a

Cited by 74 publications

(77 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A coefficient of ϩ1 represents a perfect prediction, 0 an average random prediction, and Ϫ1 the worst possible prediction. In general, MCC values greater than 0.4 are considered to be predictive in machine learning methods (Chohan et al, 2005).…”

Section: Methodsmentioning

confidence: 99%

“…For this purpose, various traditional in silico modeling methods and more recently developed nonlinear machine learning methods have been used (Chohan et al, 2005;de Graaf et al, 2005;Kriegl et al, 2005a;Yap and Chen, 2005;Fox and Kriegl, 2006;Yap et al, 2006;Eitrich et al, 2007;Terfloth et al, 2007;Zhou et al, 2007). Machine learning methods are particularly useful for data mining of large databases to discover patterns or rules to derive models for problems for which the underlying mechanism is not clear.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Classification of Cytochrome P450 1A2 Inhibitors and Noninhibitors by Machine Learning Techniques

Poongavanam

Taboureau²,

Oostenbrink³

et al. 2008

Drug Metab Dispos

View full text Add to dashboard Cite

ABSTRACT:The cytochrome P450 (P450) superfamily plays an important role in the metabolism of drug compounds, and it is therefore highly desirable to have models that can predict whether a compound interacts with a specific isoform of the P450s. In this work, we provide in silico models for classification of CYP1A2 inhibitors and noninhibitors. Training and test sets consisted of approximately 400 and 7000 compounds, respectively. Various machine learning techniques, such as binary quantitative structure activity relationship, support vector machine (SVM), random forest, kappa nearest neighbor (kNN), and decision tree methods were used to develop in silico models, based on Volsurf and Molecular Operating Environment descriptors. The best models were obtained using the SVM, random forest, and kNN methods in combination with the BestFirst variable selection method, resulting in models with 73 to 76% of accuracy on the test set prediction (Matthews correlation coefficients of 0.51 and 0.52). Finally, a decision tree model based on Lipinski's Rule-of-Five descriptors was also developed. This model predicts 67% of the compounds correctly and gives a simple and interesting insight into the issue of classification. All of the models developed in this work are fast and precise enough to be applicable for virtual screening of CYP1A2 inhibitors or noninhibitors or can be used as simple filters in the drug discovery process.Cytochromes P450 (P450s) are heme-containing enzymes found in both prokaryotes and eukaryotes, and they are involved in a wide range of cellular biotransformation functions. From a pharmaceutical perspective, the most important function is the degradation of drugs (Nebert and Russell, 2002). In general, hydrophobic compounds are converted into more hydrophilic species to facilitate excretion.The most important P450 isoforms involved in metabolism of drugs in humans are CYP1A2, CYP2A6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4. CYP1A2 constitutes 12% of the total P450 content in the liver and plays an important role in the metabolic clearance of ϳ5% of currently marketed drugs. The substrates for the CYP1A subfamily are generally characterized as neutral, flat, aromatic, and lipophilic (two to four aromatic rings) with at least one putative hydrogen bond donor (Smith et al., 1997), in agreement with the observed contacts in the recent crystal structure of CYP1A2 (Sansen et al., 2007). Examples of drugs that are CYP1A2 substrates are acetaminophen, caffeine, clozapine, haloperidol, olanzapine, propranolol, tacrine, theophylline, and zolmitriptan (drug interactions: cytochrome P450 drug interaction table, Indiana University School of Medicine, http://medicine.iupui.edu/flockhart/table.htm).In silico approaches are attractive because they can be used in an early stage of the drug discovery process and thereby reduce the number of experimental studies and improve the success rates. For this purpose, various traditional in silico modeling methods and more recently developed nonlinear machine learning methods...

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Classification of Cytochrome P450 1A2 Inhibitors and Noninhibitors by Machine Learning Techniques

Poongavanam

Taboureau²,

Oostenbrink³

et al. 2008

Drug Metab Dispos

View full text Add to dashboard Cite

show abstract

“…The maximum number of molecules in the training set is 109 compounds in the model presented by Chohan et al [2005]. Of the five models predicting CYP1A2 inhibition, a consensus model based on 4 models has the best performance with 83% correct classifications [Chohan et al, 2005]. However, the test set was biased with 18 and 231 compounds, respectively, in the two inhibition classes.…”

Section: Statistical Methods For Prediction Of Cyp Inhibitionmentioning

confidence: 99%

In silico prediction of cytochrome P450 inhibitors

Refsgaard

Jensen

Christensen

et al. 2006

Drug Development Research

View full text Add to dashboard Cite

Patients often receive several medications at the same time, and if the drugs involved compete for the same enzymes to be metabolized, it can lead to undesired effects with the risk of fatal results. Therefore, early knowledge about the cytochrome P450 (CYP) interaction potential of a drug candidate is central and in silico tools can provide such information even on virtual structures. Most of the in silico CYP information in the literature is on substrates and is based on molecular and protein modeling. However, in early screening information of CYP substrates is rarely available and sometimes only a single concentration is used in screening assays. Recently, in silico CYP modeling applying statistical tools has appeared in the literature and the aim of this review is to give an overview of published in silico prediction studies of CYP inhibition for four of the clinically most important isotypes, namely: CYP1A2, CYP2C9, CYP2D6, and CYP3A4. Furthermore, in the review, we discuss inhibition data, different descriptors and statistical methods applied for in silico prediction of CYP inhibition, and we point to promising approaches in the development of accurate in silico prediction tools of CYP inhibitors. Drug Dev. Res. 67:417-429, 2006.

show abstract

“…The interpretation of latent variables helps to understand which predictors are most involved in the prediction accuracy. In Chohan et al [312], PLSisused amongotherregression methodstopredictCytochromeP4501A2inhibition (pIC 50 ) from in-house computed descriptors accounting for topological, geometrical, and electronic features of molecules. Feature selection (according to variance, redundancy, and predictivity) was performed before PLS application.…”

Section: Technical Descriptionmentioning

confidence: 99%

“…For toxicity properties, DTs are used to predict hERG inhibition [202] and toxicity involving cytochrome P450 such as six CYP isoforms [331], 2D6 and 1A2 isoforms [193,312], or the 3A4 isoform [332].…”

Section: Decision Treesmentioning

confidence: 99%