ANN QSAR workflow for predicting the inhibition of HIV-1 reverse transcriptase by pyridinone non-nucleoside derivatives.

Abstract

Pyridinone derivatives have high potency against non-nucleoside reverse transcriptase inhibitor (NNRTI)-resistant human immunodeficiency virus type-1 strains. Quantitative structure-activity relationship (QSAR) studies on a series of pyridinone derivatives acting as NNRTIs are very important in designing the next generation of NNRTIs. Methodology & results: The QSAR models were developed using linear (single and forward stepwise) and combined nonlinear artificial neural network (ANN) approaches. ANN provided QSAR model with highly correlating values of 0.963, 0.964, 0.920 and 0.917, corresponding to the biological activity pIC50 of the training, validation, testing and all samples, respectively.The nonlinear ANN-QSAR model based on the topological polarizability, geometrical steric, hydrophobicity and substituted benzene functional group indices might be able to help for designing novel pyridinone NNRTIs.