Quantum chemical-QSAR study of some newly synthesized 1,4-dihydropyridine calcium channel blockers

A QSAR analysis was conducted on the calcium channel antagonist activity of 72 C-3 and C-5 ester-substituted 4(nitro imidazolyl) 1,4-dihydropyridine derivatives. Quantum chemical descriptors including atomic charges, electrostatic potentials, HOMO and LUMO energies, electro-negativity, electrophilicity, hardness and softness indices were calculated using ab-initio method with gaussian98 at RHF/6-21G level. Multiple linear regression was used to model the relationships between molecular descriptors and biological activity of molecules using genetic algorithm as variable selection tool (GA-MLR). A high quality multi-parametric QSAR model was obtained. The prediction ability of the resulting models was evaluated by the prediction of the activity of the prediction set compounds, which did not contribute to the model building at all. Addition of the physicochemical parameters to the electronic features of the molecules enhanced the prediction ability of the models. Artificial neural network, combined with genetic algorithm for feature selection (GA-ANN), was also employed to model nonlinear structure-activity relationships. The results showed that ANN gave more appropriate QSAR models in comparison with those obtained by MLR. The root mean square error of prediction for the best MLR and ANN models, which used 6 descriptors as predictor variables, was equal to 0.60 and 0.21, respectively. Also, the squared correlation coefficient of these models was 0.874 and 0.988, respectively.