Figure 2.

TNBC gene coexpression networks: derivation in the FUSCC cohort, functional characterization, and linkage to oncogenic mechanisms. (A) Weighted gene coexpression analysis separates genes with expression correlated across patients in the FUSCC cohort into distinct networks, as illustrated in the topological overlap matrix. Though not labeled, genes line the x and y axes, and any point in the matrix records the correlation between the corresponding genes in the pair (red reflects greater correlation). Each gene is also labeled by the color of the network to which it belongs (as derived in the coexpression analysis). Dendrograms and network colors are depicted on both axes, illustrating that the coexpression analysis in fact results in a clustering of genes into networks. (B-E) Functional characterization of networks, illustrating functionally-relevant gene sets that overlap significantly with each network for green, darkgreen, black, and brown networks, respectively: for each plot, the x-axis notes gene count within each network for respective ontology(functional) categories (labeled on the y-axis). More red signifies lower p-values. All gene pathways pictured have p < 0.05. (F) Linking gene networks to oncogenic mechanisms: protein-protein interaction network for transcription factors targeting network hub genes. Transcription factors predicted to directly bind hub genes are in pink and their interaction partners are in gray. (G) Kinase regulation of the transcription factor protein-protein interaction network: kinases are in blue, transcription factors are in pink, and their intermediate protein interaction partners are in grey. Green edges represent kinase-substrate phosphorylation interactions. Grey edges are physical protein-protein interactions.