Figure 1
From: Inferring microRNA and transcription factor regulatory networks in heterogeneous data
Method overview. The method utilises Bayesian network learning and graph search to produce a three-component regulatory network (miRNA-TF-mRNA) from multiple sources of data. Target information is used to create the initial structure representing the interactions between miRNA-mRNA, miRNA-TF, TF-miRNA, TF-TF, and TF-mRNA. For illustration, we only draw one bipartite graph in the initial structure. Expression profiles are then used in the Bayesian network learning procedure to construct the networks in each sample condition. Bootstrapping and averaging procedure is used to integrate the Bayesian networks learnt from each condition into the integrated global network. The interplays between miRNAs and TFs and the network motifs that involve at least 2 regulators are extracted from the global network and are final results.