Research Paper Volume 12, Issue 21 pp 21809—21836

Figure 3. Correlation analyses between gene cluster modules, PBRM1 mutations, and mast cell infiltration in ccRCC patients. (A) The clustering dendrogram shows different gene cluster modules that are color-coded. The dissimilarity of genes is based on the topological overlap. (B) Heatmap shows the correlation between module eigengenes and immune cell infiltration in ccRCC samples. The correlation table is color-coded. The modules in the blue box are associated with PBRM1 mutations and mast cell infiltration. (C) Analysis of the association between the 27 gene cluster modules and the 4 mutant genotypes (VHL, PBRM1, SETD2, and BAP1) in ccRCC patients. Each cell represents a module correlation co-efficient and its corresponding p-value. (D) Pathway enrichment analysis of dark orange module. Dark orange gene cluster was positive with PBRM1 mutant and mast cell infiltration. (E) Pathway enrichment analysis of the white module. White gene cluster was positive with PBRM1 mutant and mast cell infiltration. (F) Enrichment plots show upregulated FARDIN hypoxia signaling (red), MENSE hypoxia signaling (green), MIZUKAMI hypoxia signaling (green), PID-HIF1-THPATHWAY (purple), PID-HIF2-PATHWAY (blue), and other gene sets in the PBRM1^mut group of ccRCC patients. FARDIN hypoxia signaling gene set including the genes in the hypoxia signature, based on analysis of 11 neuroblastoma cell lines in hypoxia and normal oxygen conditions; MENSE hypoxia signaling gene set including hypoxia response genes up-regulated in both astrocytes and HeLa cell line; MIZUKAMI hypoxia signaling gene set including the genes up-regulated in colon cancer cells in response to hypoxia, might not be direct targets of HIF 1α; PID-HIF1-THPATHWAY gene set including the gens in HIF 1α transcription factor network; PID-HIF2-PATHWAY gene set including the gens in HIF 2α transcription factor network.