Research On Complex Disease Related Patterns Analysis Based On Multi-layer Network Modularity

The pathogenic mechanism of complex diseases is not irregular,two different diseases may have the same symptoms or similar pathogenic genes.By studying the relevant patterns of complex diseases,we can understand the pathogenesis of complex diseases in depth,and these patterns can also serve as an important basis for our drug repositioning.The traditional research on complex disease-related patterns focuses on single aspect,such as considering the relationship between two diseases only from the protein-protein interaction(PPI)network.However,other features of diseases such as the symptoms,the function of the gene,are also important features that measure the relationship between two diseases.Symptoms of the disease can reflect the internal mechanisms of activity in the organism from external performance,and it is often possible to find some interesting "complications" through similar symptoms of diseases.In addition,complex activities in the living body are accomplished by the cooperation of genes that with different functions.Therefore,understanding complex diseases from the perspective of gene function also have great significance.This article mainly understands the correlation patterns between complex diseases from a variety of different perspectives,identifies the disease similarity module by building a multi-layer disease similarity network,and we find that the disease similarity module have great significance,and this result can be applied to drug repositioning.First,this paper builds four layer disease similarity network based on different data sets.These are the disease similarity network PIDN(PPI-based Disease-disease Similarity Network)based on PPI data,the disease similarity network DSDN(Disease Symptom based Disease-disease Similarity Network)based on symptom data,the disease similarity network GODN(GO-based Disease-disease Similarity Network)based on GO(Gene Ontology)data,and the disease similarity network DODN(DO-based Disease-disease Similarity Network)based on DO(Disease Ontology)Data.Secondly,a tensor-based multi-layer network algorithm are used to identify 9 disease modules with strong correlation.These modules belong to different disease classifications.The significant analysis verifies that these modules have great significance.Through comparison with the disease similarity module identified by a single-layer network,it is find that the results obtained by this paper have higher accuracy.Finally,through analysis of KEGG(Kyoto Encyclopedia of Genes and Genomes)enrichment analysis of pathogenic genes,the internal substructure of the disease module is discovered,then we study and analyze different pathogenic mechanisms of tumor module formation.Several potential drugs for the treatment of rheumatoid arthritis are predicted through drug repositioning and molecular docking technique,and the mechanism of drug action is analyzed by us,we give a possible biological path of drug action from the perspective of cytokines.The disease similarity modules find by this paper can help us have a comprehensive understanding of complex diseases from the perspective of multi-layer networks and have very high reference value.These disease modules also provide new ideas for drug repositioning application research.