Combining DFT and QSAR studies for predicting psychotomimetic activity of substituted phenethylamines using statistical methods

The DFT-B3LYP method, with the base set 6-31G (d) was used to calculate electronic and charge descriptors. The present study was performed using principal component analysis (PCA), multiple linear regression analysis (MLR) and non-linear multiple regression analysis (MNLR) to predict unambiguous QSAR models of 46 substituted phenethylamines toward psychotomimetic activity. Results showed that the MLR and MNLR predict activity in a satisfactory manner. But among those models, we concluded that the latter one provides a better agreement between calculated and observed values of psychotomimetic activity. Also it shows very good stability towards data variations for the validation methods.