QSAR Models for Phosphoramidate Prodrugs of 2′‐Methylcytidine as Inhibitors of Hepatitis C Virus Based on PSO Boosting

Abstract

In the current study, boosting regression has been proposed to model the activities of a series of phosphoramidate prodrugs of 2'-methylcytidine as inhibitors of hepatitis C virus. The stepwise multiple linear regression and particle swarm optimization strategies are used to select descriptors which are responsible for the inhibitory activity of these compounds. As comparisons to the boosting regression method, the multiple linear regression, back-propagation neural networks, and support vector machine have also been investigated. Experimental results have shown that the boosting can drastically enhance the generalization performance of individual multiple linear regression model and the particle swarm optimization-boosting method is a well-performing technique in quantitative structure-activity relationship studies superior to support vector machine. The squared correlation coefficient and standard deviation of the best model are 0.744 and 0.438 for the training set and 0.710 and 0.748 for the test set.