The Research On Algorithm Of Identifying Gene-miRNA Functional Modules Based On The Regulatory Network

As an important regulator of eukaryotes,its regulation is crucial for biological processes and seriously affects the expression of genes..It is closely related to a variety of cancers and the study of miRNAs plays an important role in the clinical treatment and drug design of cancer.At the same time,There is exist synergistic and cooperative relations between miRNAs,and they act on the genes and affect the gene expression.miRNA co-operation with miRNA,and its regulation of the target gene,together form the regulatory network of gene-miRNA function module.The functional modules play a role in the production and development of biological cell functions.Therefore,how to identify the gene-miRNA function modules in the regulatory network plays an important biological role in exploring the disease life mechanism and the causes.In this regard,this paper presents two different miRNA regulatory network gene-miRNA function module recognition algorithm,the main research work are:At present,there are many recognition modules of gene-miRNA functional modules which can not recognize overlapping functional modules.At the same time,the research shows that the fusion multi-network can effectively improve the recognition accuracy of the algorithm.This paper proposes a new gene-miRNA functional module recognition algorithm MCLM based on improved Markov clustering.The algorithm firstly construct miRNA functional similarity network by using multi-network fusion to miRNA-miRNA weighting;Then for the acquired miRNA functional similarity network,using improved Markov clustering algorithm to obtain miRNA clusters;Finally,for each miRNA cluster and for each miRNA,Using the random forest classification algorithm to rank the target genes and then selectting the top-ranked genes to join the corresponding miRNA clusters,eventually identifying the gene-miRNA functional modules on regulatory networks.Experiments show that gene-miRNA functional modules recognized by MCLM are involved in a large number of biological processes and have important biological significance.Although the above algorithm can effectively identify overlapping functional modules,it only generalizes the gene-miRNA functional modules related to certain diseases.In order to further identify functional modules related to specific subtypes of diseases,a bi-clustering and Bayesian network-based gene-miRNA functional module recognition algorithm BBM is proposed.The algorithm firstly uses the double clustering algorithm(SAMBA)to build the gene-sample module;Then for each gene-sample module,using the average PCC value of the gene in the module and PPI data for module expansion,and filter out non-compliant modules by calculating the salient p-value of the module;Finally,for each GSM,the miRNAs are ranked by calculating the correlation between each miRNA and the gene in the module,and then select the top miRNAs to join the module and finally identify the gene-miRNA functional modules on the regulatory network.Experiments show that the BBM algorithm can effectively identify gene-miRNA functional modules of specific disease subtypes with important biological significance,it plays a significant role in understanding the regulation network and revealing the mechanism of human complex diseases.