QSAR analysis of some l-(3,3-diphenylpropyl)-piperidinyl amides and ureas as CCR5 inhibitors using genetic algorithm-least square support vector machine

摘要

Quantitative relationships between structures of sixty-seven of l-(3,3-diphenylpropyl)-piperidinyl amides and ureas and their CCR5 inhibitory activities were investigated by GA-MLR and GA-LS-SVM. As a preliminary step, selection of relevant descriptors by genetic algorithm was performed and then linear dependence was established by MLR. Selected descriptors by genetic algorithm were also used as input of LS-SVM model. Comparison of the GA-MLR and GA-LS-SVM models discloses superiority of the GA-LS-SVM over the GA-MLR model. Obtained results show that there is a nonlinear relationship between the pIC_50s and the calculated structural descriptors of the l-(3,3-diphenylpropyl)-piperidinyl amides and ureas. The accuracy and predictability of the proposed models were illustrated using various validation techniques.