XPredRBR: Accurate and Fast Prediction of RNA-Binding Residues in Proteins Using eXtreme Gradient Boosting

A variety of studies have shown that protein-RNA interactions play a vital role in many fundamental cellular processes, such as protein synthesis, mRNA processing, mRNA assembly, ribosome function and eukaryotic spliceosomes. Identification of RNA-binding residues (RBR) in proteins is an key step to understand the mutual recognition mechanism underlying the protein-RNA interactions. In this paper, we proposed a novel method, XPredRBR, to predict the RNA binding residues in proteins, by exploiting the eXtreme Gradient Boosting (XGBoost) algorithm. Two types of new predictive features derived from residue interaction network and solvent exposures are combined with conventional sequence features and structural neighborhood features to predict RBR. We carried out empirical experiments on two datasets to demonstrate the performance of the proposed method. By 10-fold cross-validations, our method achieved the accuracy of 0.861, sensitivity of 0.872, MCC of 0.584 and AUC of 0.941 on the RBP170 dataset. On another independent test set RBP101, XPredRBR outperformed three traditional classifiers and seven existing RNA-binding residue methods. A case study on the chain E of 3PLA protein illustrated XPredRBR effectively identified most RNA-binding and non RNA-binding sites. Furthermore, XPredRBR is much faster than our previous method PredRBR. These experimental results show that our proposed method achieves state-of-the-art performance in predicting RNA-binding residues in proteins.