TY - JOUR

T1 - Classification of cytochrome P450 1A2 inhibitors and noninhibitors by machine learning techniques

AU - Poongavanam, Vasanthanathan

AU - Taboureau, Olivier

AU - Oostenbrink, Chris

AU - Vermeulen, Nico P E

AU - Olsen, Lars

AU - Jørgensen, Flemming Steen

PY - 2009

Y1 - 2009

N2 - The cytochrome P450 (CYP) 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 CYPs. In this work, we provide in silico models for classification of CYP1A2 inhibitors and non-inhibitors. Training and test sets consisted of about 400 and 7000 compounds, respectively. Various machine learning techniques, like binary QSAR, support vector machine (SVM), random forest, kappa nearest neighbors (kNN), and decision tree methods were used to develop in silico models, based on Volsurf and MOE 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 - 76 % of accuracy on the test set prediction (Matthews Correlation Coefficient 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 the developed models in this work are fast and precise enough to be applicable for virtual screening of CYP1A2 inhibitors or non-inhibitors, or can be used as simple filters in the drug discovery process.

AB - The cytochrome P450 (CYP) 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 CYPs. In this work, we provide in silico models for classification of CYP1A2 inhibitors and non-inhibitors. Training and test sets consisted of about 400 and 7000 compounds, respectively. Various machine learning techniques, like binary QSAR, support vector machine (SVM), random forest, kappa nearest neighbors (kNN), and decision tree methods were used to develop in silico models, based on Volsurf and MOE 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 - 76 % of accuracy on the test set prediction (Matthews Correlation Coefficient 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 the developed models in this work are fast and precise enough to be applicable for virtual screening of CYP1A2 inhibitors or non-inhibitors, or can be used as simple filters in the drug discovery process.

KW - Former Faculty of Pharmaceutical Sciences

U2 - 10.1124/dmd.108.023507

DO - 10.1124/dmd.108.023507

M3 - Journal article

C2 - 19056915

VL - 37

SP - 658

EP - 664

JO - Drug Metabolism and Disposition

JF - Drug Metabolism and Disposition

SN - 0090-9556

IS - 3

ER -