Prediction of Metabolic Stability of Xenobiotics by the Pass and Gusar Programs

Abstract

Metabolic stability determines the susceptibility of compounds to biotransformation on the body. It is characterized by such parameters as half-life (T_1/2) and clearance (CL). In vitro systems based on cells or subcellular fractions (mainly liver microsomal enzymes) are consider as models of processes occurring in a living organism and are used for metabolic stability assessment. The data obtained from the experiments are used to build QSAR models. Based on the freely available database ChEMBL v.27, we collected more than 8000 records containing the structures of compounds and their clearance and/or half-life time values obtained on human liver microsomes. GUSAR (General Unrestricted Structure-Activity Relationships) and PASS (Prediction of Activity Spectra for Substances) software were used to create quantitative and qualitative models based on the collected data. A 5-fold cross-validation procedure was used to the model assessments. Thresholds T_1/2 = 30 min and CL = 20 mL/min/kg were chosen to distinguish between stable and unstable molecules. The accuracy of the models changes from 0.5 (calculated using 5-fold cross-validation on quantitative models for predicting the half-life) to 0.96 (calculated using 5-fold cross-validation on classification models for predicting the clearance).