Figure 1.

Workflow of module identification. 1) Profile expression of both miRNAs and mRNAs in the same set of samples using microarrays. 2) Calculate miRNA-mRNA correlation matrix based on the similarities in the expressions across samples. 3) Estimate false detection rates for a series of thresholds, and choose one based on the desired false detection rate to convert the correlation matrix into a binary miRNA-mRNA correlation network. 4) Construct a miRNA-mRNA regulatory network by combining the constructed miRNA-mRNA correlation network and the corresponding miRNA-target matrix. 5) Represent the regulatory network as a bipartite graph. 6) Enumerate all maximal bicliques as candidate regulatory modules, and post-process candidate modules, including the assessment of both the statistical significances, and differential expressions of target mRNAs between HCV+ and HCV-.

Peng et al. BMC Genomics 2009 10:373   doi:10.1186/1471-2164-10-373
Download authors' original image