Predicting Protein - RNA Binding Sites Using Sequence Statistical Feature of Amino Acids

Protein and RNA interactions play essential roles in a number of biological regulatory mechanisms. Effective identifications of the binding interfaces can help understand the interactions. In this paper, we take statistical information into account, mainly the singlet propensity and doublet propensity, and add the two propensities with the sequence information, to predict the interfaces using machine learning method. Results show that adding statistical characters can improve the prediction precision, especially the doublet propensity. Besides, we constructed three data sets based on the protein-RNA complex function, and find out that different complex function data sets show significant differences in prediction precision.