A number of quantitative structure-activity relationship (QSAR) models has been developed utilizing theoretical molecular descriptors such as topological, geometrical and functional group indices calculated solely from the structures of 80 set of synthesized 1-(2-hydroxyethoxy-methyl)-6-(phenylthio) thymine (HEPT) derivatives which are non nucleoside reverse transcriptase inhibitors shown to have both potent anti-HIV activity and inhibit HIV-1 at nanomolar concentration. The QSAR models are generated by using multiple linear regression method. Impacts of such computed structural descriptors towards reverse transcriptase inhibitory activities of these compounds were analysed by stepwise forward backward variable selections.The developed QSAR training models are statistically validated. Topological indices can contribute the maximum impact on biological activity obtained in terms of model quality such as R2 = 0.903, QLoo2 = 0.850, Rpred2 = 0.620 respectively. Whereas geometrical and functional group descriptors have produced almost similar influences on the activity although external predictability using functional group indices is higher than the geometrical descriptors. Most significant descriptors having crucial influences on the reverse transcriptase inhibition include path/walk 4 - Randic shape index (PW4), topological charge indices and Eigen value from edge adjacency matrix weighted by resonance integrals, information content and total information index on atomic composition, quadrupole x component value / weighted by polarizability (QXXp), Folding degree index (FDI) and sum of geometrical distances between S..Cl at 6-phenyl thio moiety of the HEPT derivatives, number of acceptor atoms for H-bonds (N,O,F), number of un substituted benzene C(sp2) and CHR3 (C-003). The derived significant models in such chemo metric descriptors may be used to design and synthesize new potential compounds in this series and the studies in this direction would focus designing of potent anti HIV-1 HEPT compounds considering the predicted essential structural features predicted by our developed models.
