Abstract

Predicting the interaction between drugs and target proteins is one of the most important tasks for bioinformatics. Selecting suitable features for drug-target interaction prediction is an effective way for achieving this important goal. In this article, the molecular descriptors information of drugs combined with several important biological features of nuclear receptor (NR) proteins were first utilized for feature selection. Then, the incremental learning algorithm with support vector machines was used to obtain the optimal feature subset for drug-target interaction prediction. The results demonstrate that the feature selection method can lead to promising improvement on prediction accuracy of this target family.

Keywords: Drug-target interaction, Molecular descriptors, Feature selection.