The Study Of Computational Network Method To Predict RNA Binding Sites

RNA is one of the basic units in the cell,it usually forms the complexes with protein or ligand to perform biological processes,such as regulation,catalysis etc.However,it is very difficult to determine the complex’^experimental structure due to the RNA flexibility and technology limitations,that hinders the complex’s functional exploration.Therefore,the theoretical RNA binding site study is able to help us to understand the relationship between structure and function,and explain many biological mechanisms.In this paper,we developed a network strategy algorithm RBind to predict RNA binding sites based on RNA tertiary structures.First,the RNA tertiary structures were transformed into a network.In the network,the nucleotides are nodes,and two nonconsecutive nucleotides in a sequence of RNA are connected by an edge if the shortest distance of pair of heavy atoms from different nucleotides is less than 8 A.Second,the closeness and degree values of each node in RNA network were calculated.Third,we identified the RNA binding sites based on a cutoff value calculated by the combination of degree and closeness.Furthermore,we tested RBind in RNA-ligand and RNA-protein testing datasets.The results showed that the accuracy values(0.82 in RNA-ligand dataset and 0.63 in RNA-protein dataset)are higher than current available RNA binding sites prediction algorithms.We also predicted the binding sites of modeled RNA structure,the results showed that RBind has a reliable prediction if the structure model was in a reasonable range of accuracy.And the direct coupling analysis(DCA)for nucleotide-nucleotide showed that binding sites may coevolve to maintain structure and functions.This provided the enlightenment in RNA binding sites prediction.The method of binding sites prediction is helpful for computer aided drug design(CADD)and the prediction of complex structures.We also have built a corresponding website for RBind.All the codes and datasets can be downloaded from https://zhaolab.com.cn/RBind.