QSAR studies of imidazopyridine derivatives as Et-PKG inhibitors using the PSO-SVM approach

In this work, the multiple linear regression (MLR) and support vector machine (SVM) methods were applied for modeling and predicting the inhibitory activity of imidazopyridine derivatives with two-dimensional (2D) autocorrelation descriptors calculated from the molecular structure alone for the first time. We define the new objective function as fitness, and it can guide the particle swarm optimization (PSO) method to select important descriptors which are responsible for the inhibitory activity of these compounds. The square of the correlation coefficient (R (2) = 0.897), the square of correlation coefficients of the test set (R-ext(2) - 0.660), and the obtained statistical parameter of the calibrating set in the PSO-SVM model was 0.743, which demonstrated the reliability of the model. The PSO-SVM model is superior over the PSO-MLR method in the dataset with imidazopyridine derivatives as Et-PKG inhibitors. Our best quantitative structure-activity relationship model illustrates the importance of an adequate distribution of atomic properties represented in topological frames and reveals atomic masses, van der Waals volumes, Sanderson electronegativities, and polarizabilities to be the most influential atomic properties in the structures of the imidazopyridine derivatives.