Fig. 2

Master Regulator Analysis Pipeline for a pancancer invasiveness phenotype. A A sample RNA-Seq matrix of genes vs samples where rows represents genes and columns represent tumor samples. B Reverse-engineered gene regulatory network using RGBM technique. Each big blog represents a transcription factor (TFs) and the small dots represent target genes. The regulatory network is divided into communities (color-coded) using Louvain clustering algorithm. C An example of the adjacency list corresponding to the inferred GRN as well as the correlation matrix between the TFs and target genes based on the RNA-seq matrix. D The sample-TF activity matrix as estimated by Eq. 2 where rows are TFs and columns are tumor samples. E Significantly enriched TRs (FDR-adjusted P-value < 0.05) referred to as master regulators (MRs) identified by using the FGSEA method. F The sample-MR activity matrix extracted from sample-TF activity matrix. The activity matrix has block diagonal structure with some MRs having high activity in INV-H but low activity in INV-L while other MRs vice versa activity profiles