Figure 3

DCIM derived contexts and related differentially co-expressed genes. DCIM was used to identify contexts and differentially co-expressed genes in a breast cancer dataset [37]A) Hierarchical clusterings of patients based on differential co-expression PPPs and the heatmap of 200 most differentially co-expressed genes between two contexts marked (1) and (2). The bottom panel shows the average expression profile for the three global gene clusters marked in the heatmap with corresponding color sidebars. The right-hand panel shows significantly enriched functional categories for these genes as determined by CLEAN [55] where red indicates the corresponding cluster is significantly enriched by the category and green indicates no significant cluster enrichment. Complete CLEAN results for all possible gene clusters can be interactively browsed using the FTreeView software at the support website http://eh3.uc.edu/gimm/dcim. B) Empirical distribution of all pairwise gene-gene correlation coefficients (Pearson correlation) for the 154 genes marked by the left sidebar in A. The top right plot shows correlations for 154 randomly selected genes in the same two contexts while the bottom right plot shows correlations for the same genes but with randomly reassigned context labels.