Predicting NcRNA-protein Interactions With Graph Neural Networks

Noncoding RNAs(nc RNAs)are made by transcription of non-coding genes.They play crucial roles in many biological processes.One of the important research directions of nc RNA is to study the interaction between nc RNA and protein.Because identifying nc RNA-protein interactions by experimental methods is expensive and time-consuming,many computational methods are rapidly evolving as alternatives.Graph neural network(GNN)is a recently developed deep learning algorithm,and GNN has been used in link prediction of complex networks.This research constructed a GNN-based method,which is called Noncoding RNA–Protein Interaction prediction using Graph Neural Networks(NPI-GNN),to predict nc RNA-protein interactions.This research collected five benchmarking datasets for predicting nc RNA-protein interactions.Based on these datasets,this research evaluated and compared the prediction performances of existing methods and NPI-GNN.The NPI-GNN method achieved comparable performance with state-ofthe-art methods in a 5-fold cross-validation.In addition,it is capable of predicting novel interactions based on network information and sequence information,helping future nc RNA research.In this study,the effects of different datasets on NPI-GNN are quantitatively analyzed,so as to realize the advance estimation of NPI-GNN performance based on datasets.This research also found that insufficient sequence information does not affect the NPI-GNN prediction performance much,which makes NPI-GNN more robust than other methods.As far as we can tell,NPI-GNN is the first end-to-end GNN predictor for predicting nc RNA-protein interactions.All benchmarking datasets in this work and all source codes of the NPI-GNN method have been deposited with documents in a Git Hub repo(https://github.com/Ashui RUA/NPI-GNN).