PD-175069

A quantitative structure-activity relationship (QSAR) analysis was performed on a data set of 104 molecules showing N-type calcium channel blocking activity. Several types of descriptors, including electrotopological, structural, thermodynamics and ADMET, were used to derive a quantitative relationship between N-type calcium channel blocking activity and structural properties. The genetic algorithm-based genetic function approximation (GFA) method of variable selection was used to generate the 2D-QSAR model. The model was established on a training set of 83 molecules, and validated by a test set of 21 molecules. The model was developed using five information-rich descriptors--Atype_C_24, Atype_N_68, Rotlbonds, S_sssN, and ADME_Solubility--playing an important role in determining N-type calcium channel blocking activity. For the best QSAR model (model 4), the statistics were r (2) = 0.798; q (2) = 0.769; n = 83 for the training set. This model was further validated using the leave-one-out (LOO) cross-validation approach, Fischer statistics (F), Y-randomisation test, and predictions based on the test data set. The resulting descriptors produced by QSAR model 4 were used to identify physico-chemical features relevant to N-type calcium channel blocking activity.