Quantitative structure–activity relationship study of acyl ureas as inhibitors of human liver glycogen phosphorylase using least squares support vector machines

Abstract

An effective quantitative structure–activity relationship (QSAR) model of a series of acyl ureas as inhibitors of human liver glycogen phosphorylase a (hlGPa), was built using a modified algorithm of support vector machine (SVM), least squares support vector machines (LS-SVMs). Each compound was depicted by structural descriptors that encode constitutional, topological, geometrical, electrostatic and quantum-chemical features. The Heuristic Method (HM) was used to search the feature space and select the structural descriptors responsible for activity. The LS-SVMs and multiple linear regression (MLR) methods were performed to build QSAR models. The LS-SVMs model gives better results with the predicted correlation coefficient (R) 0.899 and mean-square errors (MSE) 0.148 for the test set, as well as that 0.88 and 0.174 in the MLR model. The prediction results indicate that LS-SVMs is a potential method in QSAR study and can be used as a tool of drug screening.