Molecular docking and receptor-based QSAR studies on inhibitors of PARP-1 as a potential target for cancer therapy

One of the important molecular targets in antitumor drug discovery is Poly ADP-ribose polymerase-1 (PARP-1) enzyme. PARP-1 is one of the key members of DNA repair enzymes that is responsible for the repair of DNA single-strand breaks. Inhibition of PARP-1 has been demonstrated to be a promising strategy to selectively kill tumor cells by targeting DNA repair pathway [1-3]. In the present report, efficient linear and non-linear methods including multiple linear regression (MLR) and support vector machine (SVM) used to develop and establish quantitative structure-activity relationship (QSAR) models capable of predicting half-maximal inhibitory concentration (IC50) values of phthalazinones derivatives as effective PARP-1 enzyme inhibitors. The data set consisted of 77 chemicals that were divided into two subsets of training and test set through Kennard-Stone algorithm. Forward MLR method was used for reduction of the number of descriptors and feature selection. In this method, the variables are added to MLR model step by step noting the effect on the R value of the model. By applying this method eight descriptors selected and used as input variables in developing of MLR and SVM models. Among these models, SVM model with eight selected molecular descriptors was showed better statistical parameters (R train = 0.947, RMSE train = 0.324, Q2 = 0.728, R test = 0.921, RMSE test = 0.393). As a complementary method, molecular docking was used to study the mechanism of interactions between ligands and protein in the active site. Here, all derivatives were docked and the differences in the structures of compounds are exhibited in docking study.