Abstract

Quantitative structure protein binding relationships (QSPBR) have been derived for the binding of nine antipsychotic phenothiazine derivatives to human alpha-1-acid glycoprotein (AGP). Ninety eight topological indices for the phenothiazine derivatives were calculated using software Polly 2.3. Bivariate correlation matrix showed that most of the statistically significant correlations were obtained for topochemical indices. Four variables, complementary information contents, CICo, CIC1 and CIC2 and chain connectivity index, 6Xch, with correlations significant up to 0.01 level were considered to be the predictors of the model. Linear, multiple linear and non-linear regression analysis was performed on the retained variables. The complementary information content for 2nd order neighbourhood of vertices in a hydrogen filled graph, CIC2, was found to be the best single parameter that can be used in linear or quadratic fit to predict the binding affinities of phenothiazines to AGP.