In the present work, quantitative structure activity relationship studies were performed to explore the structural and physicochemical requirements of 1-[(2-hydroxyethoxy)methyl]-6-(phenylthio)thymine (HEPT) derivatives for anti- HIV activity. QSAR models have been developed using steric, electronic and thermodynamic descriptors. Statistical techniques like genetic function approximation-multiple linear regression (GFA-MLR) as the data preprocessing step were applied to identify the structural and physicochemical requirements for anti-HIV activity. The generated equations were statistically validated using leave-one-out technique and the best models were also subjected to leave-5-out cross-validation. The quality of fit and predictive ability of equations obtained from GFA-MLR is of acceptable statistical range (explained variance of 91.74%, while predicted variance of 74.14%). The robustness of the best models was checked by Y–randomization test and identified as good predictive models. The coefficient of ALogP, ATSm5 and CrippenLogP shows that the activity increases with increase in ALogP, ATSm5 and Crippen LogP of molecules. The coefficient of C2SP3, VPC-4, SsI, ETA_AlphaP, ETA_Epsilon_1, nAtomP, Petitjean Number and Wlambda2.unity shows that the activity decreases with increase in volume and Wlambda1.polar of the molecules is detrimental to activity. The information generated from the present study may be useful in the design of more potent HEPT derivatives as anti HIV agents.
