The Research On Algorithm Of Identifying Key Regulators Based On The TF-miRNA Co-regulatory Network

In eukaryotes,there are two kinds of important regulatory factors:transcription factor(Transcriptional Factor,TF)and microRNA(miRNA),they regulate the expression of target genes at transcriptional and post-transcriptional levels,respectively,and this regulatory role throughout a variety of biological activities and disease process.On this basis,the study found that there is exist a wide range of interaction and cooperation regulation between transcription factors and miRNAs,they form a complex co-regulatory network,it contains richer biological information than a single network.The study of key regulators in co-regulatory network is one of the important ways to understand the regulation and regulate mechanism of co-regulation network,therefore,how to identify the key regulators in the co-regulatory network becomes the key to understanding the mechanism of gene co-regulation network and revealing the mechanism of human complex diseases.In this paper,we propose two different key regulators recognition algorithms in co-regulatory networks,the main research work are as follow:Because of the existing key regulatory recognition algorithms,there is little consideration for the co-regulatory effects of TF and miRNA on genes,a new key regulators recognition algorithm named co-BOTLM for co-regulatory network based on linear model is proposed.The algorithm first integrates miRNA-gene,TF-gene and gene-gene interaction data and expression data,to construct a linear model to predict the expression of known gene,and fully consider the cooperative control relationship between the regulators in the co-regulatory network;And then measure the influence of regulators on the gene of the known disease according to change the expression level of the disease gene,so as to obtain the activity value of miRNA,TF and gene in the network more accurately;Finally,the linear model is transformed into an optimization problem by minimizing the difference between the predicted expression and the true expression value of the known disease gene,at last,the key regulators in the co-regulatory network are identified.Experiments show that the key control subgroup identified by the co-BOTLM algorithm are involved in a large number of biological processes and have important biological significance.In order to reduce the time complexity of the algorithm and further improve the recognition accuracy of the algorithm,a new key regulators recognition algorithm named co-LRM based on the binary logistic regression model is proposed.The algorithm regards the key control sub-identification problem as a two-class problem,first assigns a priori label to each regulators according to a prior information to construct a binary logistic regression model;And then the eigenvector is obtained by multi-network integration to solve the logistic regression model,and the posterior probability of each regulator was calculated;Finally,the regulators were ranked according to the posterior probability,and eventually identify the key regulators on the regulatory network.Through experiments show,co-LRM algorithm can effectively identify key regulators with significant biological significance.At the same time,compared to the algorithm co-BOTLM,Running time of algorithm co-LRM is relatively low.