Figure 1.

Workflow. STEP 1: To detect interacting transcription factor (TF) and target gene (TG) modules, respectively, prior protein–protein interaction (PPI) network was instantiated with gene expression data and community detection algorithm was used to detect condition-specific TF and TG modules. STEP 2: To get putative TF–TG relations, we conducted projection from a TF module to a TG module through public gene regulatory network (GRN). This process is conducted for every possible TF–TG module pair. STEP 3: Utilizing kernel canonical correlation analysis (CCA), we constructed condition-specific GRN that detects multiple-to-multiple regulatory relationships between TFs and TGs.