A Learning Method for Predicting the Binding Strength in RNA-Protein Complexes

Abstract

Hydrogen bond and van der Waals interactions between protein and RNA are important. We have developed a set of algorithms for predicting RNA-Protein binding strength by analyzing hydrogen bond and van der Waals interactions between protein and RNA. Firstly, we must identify the RNA-Protein binding sites. In this study, we use features including Pseudo Position-Specific Score Matrix (PsePSSM) computed by PSI-BLAST and Dipeptide Composition (DC) as feature vectors. Then, the classifier is employed to identify the residues that interact with RNA in RNA-binding protein. Then, take into account the number of amino acids hydrogen bonding and van der Waals forces to any nucleotide, the binding strength is calculated. Finally, fuzzy sets method is adopted to predict the binding strength is strong or weak. Our experiments show that the above methods are used effectively to deal with this complicated problem of predicting RNA-protein binding strength.