Prioritization Of Candidate Disease Genes Based On Topological Similarity And Optimized PPI Network

Many diseases,like diabetes,cancer,cardiovascular disease,are closely related to genes.These diseases are becoming more and more popular in both developed and developing countries.Illustrating the relationship between gene and human genetic disease has become one of the most emerging and important topic in current systematic biology.Researches show that genes that cause the same disease or diseases with same phenotype tend to be interact with each other in the gene interaction network.Therefore,prediction method based on protein-protein interaction network(PPI network)are becoming popular.However,some of the interaction between proteins are weak or missing,thus the interaction relationship between genes are not well presented and preventing the improvement of the prediction accuracy.To overcome this defect,this pape In order to overcome this shortcoming,this paper supplements the PPI network and the random walk algorithm based on the introduction of gene ontology data,phenotypic similarity data and gene-phenotype relationship data,and improves the random walk algorithm in the protein interaction network.We firstly complements the PPI network with the Gene Ontology(GO)data to generate a weighted PPI network.Secondly,a gene-phenotype heterogeneous networks is generated according to the phenotype similarity and gene-phenotype network,and the RWR algorithm is applied in this network which the known disease genes are set as the seed nodes.In this stage,several candidate disease genes will be selected.Lastly,these candidate disease genes are set as the seed nodes with the weighted obtained in the second stage and the RWR algorithm is applied to the weighted PPI network to obtain the final prediction results.This approach essentially addresses the problem of data incompleteness by introducing additional bioinformatics data to complement the protein interaction network.In summary,this paper proposed a disease genes prediction method based on weighted PPI network optimized by Gene Ontology data and RWR algorithm.This method sets the initial weight of the seed nodes according to the gene-phenotype network.The prediction accuracy is improved by introducing the Gene Ontology and gene-phenotype data to the PPI network and the effectiveness of this method is illustrated.