2D autocorrelation modelling of the anti-HIV HEPT analogues using multiple linear regression approaches

A quantitative structure-anti HIV-1 activity relationship study has been applied in a series of 1-[2-hydroxyethoxy-methyl]-6-(phenylthio) thymine] analogues acting as non-nucleoside reverse transcriptase inhibitors. The relevant 2D autocorrelation descriptors for deriving a quantitative relation between the anti-HIV activity and structural properties were selected by the multiple linear regression approach. Analysis of the resulting model revealed a correlation coefficient and a root mean square error of 0.859 and 0.503, respectively. The predictive ability of the model indicates that this model can be used for virtual library screening of databases for novel potent anti-HIV agents.