Predicting West Nile Virus-human Protein Interactions Based On Structural Similarity

West Nile virus(West Nile virus,WNV)is a parasitic virus of flavivirus.In recent years,it has posed a great threat to human's health in the world due to its outbreak.Studying the protein-protein interactions(PPIs)between WNV and human can contribute to explaining the pathogenesis of WNV and the immune response mechanism of human at the molecular level.Because the PPIs experimental data of WNV-human are relatively scarce,the development of computer methods are expected to make up the defects of experimental methods.The existing research are mainly based on the protein sequence information to predict cross-species PPIs,however,the structure of the protein is far more conservative than sequence,so it is possible to make full use of protein structure information to improve the predictive reliability of WNV-human PPIs.In order to construct WNV-human PPIs network,our study used protein structure similarity to identify human interacting partners which interact with WNV.Firstly,we evaluated structural similarity of WNV proteins and human ones by structural comparison.Human PPIs database was used as template library to obtain WNV-human PPIs.After screening these interacting pairs by subcellular co-localization,a network of 3,347 interactions was constructed,which included 6 WNV proteins and 1,970 human proteins.By analyzing the evolution rate of the target proteins and their network topological properties in the host interaction network,we found that these proteins are more conserved at the evolutionary level than the non-target ones,and they tend to be the central and bottleneck proteins in the human PPIs network.Triplicate analysis showed that the target proteins are adjacent to each other and have co-expression characteristics in the intracellular network,suggesting that these proteins may have similar biological functions.In this study,we carried out GO function annotation and KEGG enrichment analysis for these targets,and found that they are mainly involved in many biological processes,such as virus process,transcription regulation,cell adhesion and so on.What's more,they are mainly related to neurotrophin signaling pathway.This result is helpful in elucidating the pathogenesis of WNV and the molecular mechanism of host immune response.In addition,899 common targets and 267 WNV specific targets were predicted by comparing the network between WNV-human and DENV2(Dengeu virus II,DENV2)-human.By comparing the properties of sequences,network topology and rate of evolution of these two kinds of proteins,we found that the common target may be a protein with longer sequence and higher hydrophobicity,and they tend to be more conserved and as key nodes in the human network.Finally,34 potential anti-WNV targets were identified by combining the network topological features and existing drug target information,11 of these targets were reported to be associated with WNV infection.In addition,291 small molecule drugs were screened from the drug database.And then we constructed an interaction network between targets and drugs,which is expected to provide a clue for the treatment of WNV-induced diseases.