Quantitative Structure Activity Relationship (QSAR) of Methylated Polyphenol Derivatives as Permeability Glycoprotein (P-gp) Inhibitors: A Comparison of Different Training and Test Set Selection Methods

Abstract

Background: P-glycoprotein (p-gp) is one of the membrane transporter protein belong to the ATP-binding cassette which can efflux drugs to the out of the cell and cause drug resistance. Therefore, designing of new compounds with p-gp inhibitory activity can reduce drug resistance. Objective: Our aim is to introduce quantitative structure activity relationship (QSAR) models for predicting the p-gp inhibitory activity of the methylated polyphenol derivatives. Methods: Structure and activity of 52 compounds were obtained from the literature. Structure of the molecules were optimized using Hyperchem software, and molecular descriptors were calculated by the Dragon software. For external validation of the QSAR models, the data split to training and test sets using random sampling and rational methods (activity sampling and Kennard-Stone algorithm). The QSAR models were established by using both linear methods, i.e., multiple linear regression (MLR) and non-linear methods, i.e., artificial neural networks (ANN) and support vector machine (SVM). Results: Non-linear models and rational training and test set selection methods can introduce better results for predicting the activity. Conclusion: The developed QSAR models were able to predict the p-gp inhibitory activity of the studied compounds with good accuracy.