In silico prediction of cellular permeability of diverse chemicals using qualitative and quantitative SAR modeling approaches

Abstract

The permeability of the molecules through the cultured Caco-2 cells is an established in vitro method for the assessment of the absorption of oral drugs. The computational approach for predicting the cellular permeability of molecules may potentiate the screening of new drugs. In this study, gradient boosted tree (GBT) approach based qualitative and quantitative structure–activity relationship (SAR) models have been established for binary classification (moderate–poor and highly permeable) and permeability prediction of molecules using the Caco-2 cell dataset. The structural diversity of the chemicals and nonlinear structure in the considered data were tested by the similarity index and Brock–Dechert–Scheinkman statistics. The external predictive power of the developed SAR models was evaluated through the internal and external validation procedures recommended in QSAR literature. In complete data, the qualitative SAR model rendered classification accuracy of 99.26%, while the quantitative SAR model yielded a correlation (R2) of 0.917 between the measured and predicted permeability values with the mean squared error (MSE) of 0.08. The results suggest for the appropriateness of the developed SAR models to reliably predict the cellular permeability of diverse chemicals in Caco-2 cells and can be useful tools for initial screening of molecules in the drug development process.