Predicting Disease Genes Algorithm Based On Network Structure Reasoning And Topological Similarity

Now days,researchers found that almost all diseases are associated with genes,but how to find quickly the disease pathogenicity of genes is a big challenge for all researchers.Researchers use previously traditional methods like biological experiment to predict disease genes,such as linkage analysis and association studies.In recent years,researchers have greater attention to calculation methods of bioinformatics,and try to solve this problem through a network-based approach.However,most of the methods only use the local network information in the reasoning process,and they are limited to the reasoning of single gene association.These methods have a little consideration of the network topology similarity in disease and gene-related networks.In this article,203 diseases data in the OMIM database are used as the experimental dataset to predict the pathogenic genes by combining the relevant knowledge in the field of bioinformatics and data mining.The main contents of this paper are as follows:1.In order to predict the disease genes,we use the bipartite network structure reasoning algorithm for the first time.Firstly,we construct a bipartite graph network between diseases and genes;Then initialize genetic resources to diseases by using the principle of material thermal diffusion;Finally,disease resource is spreading to the genes.After the above operation,the resource vector of the candidate gene will be acquired.And the potential disease genes are predicted by sorting the vector value.In this article,203 diseases data in the OMIM dataset are used as the final experimental data,and the validity of the method is proved by leaving a cross validation experiment.2.Aiming at the reasoning algorithm based on bipartite graph network structure,we propose a method based on network topology similarity,which called INBI algorithm.Firstly,we compute the topological similarity between the topological properties of the genetic network and the disease network and use the Gaussian kernel function to compute it.Secondly,construct the disease adjacency matrix and the gene adjacency matrix.Finally,we calculate the correlation scores of to predict the potential pathogenic genes.After comparing the experimental results of the analysis,we can conclude that the performance of INBI algorithm is better.