Abstract

The DFT-B3LYP method, with the basis set 6-31G (d, p), was employed to calculate some quantum chemical descriptors of 33 biphenyl imidazole derivatives as bombesin receptor subtype-3 agonists. The descriptors were then employed to establish a quantitative structure activity relationship (QSAR) using combination of principal component analysis (PCA) and radial basis function neural network (RBF). The statistical results indicate that the correlation coefficient (R²) and cross validation using leave-one-out were 0.968 and 0.963, respectively. To validate the predictive power of the resulting model, external validation was carried out on the test set. The results show that the PCA-RBF model has not only favorable estimation stability but also good prediction power. Furthermore, it can be concluded that the agonist activity of studied compounds toward the bombesin receptor subtype-3 depends on the electronic distribution.

Keywords: Bombesin receptor subtype-3 agonists, Density functional theory, Radial basis function neural network, QSAR, Biphenyl imidazole derivatives, Nonlinear relationship.