Characterized by immune-related biology, including claudin-low, C1, C2, HCC Chiang’s inflammation, lung SCC secretory and type I NPC, were overrepresented in the group K1. On the other hand, luminal A, C5, S3, C3 and type II NPC were overrepresented in group K2. An obvious difference in immune-related pathway activities between these two groups was found. We found that both Hoshida’s S1 and Chiang’s Proliferation were enriched in group K1 while Hoshida’s S3 and Chiang’s CTNNB1 were enriched in group K2. This is consistent with the observation that both Hoshida’s S1 was significantly enriched with gene signature of Chiang’s Proliferation while Hoshida’s S3 was significantly enriched with gene signature of Chiang’s CTNNB1. Thus, subtypes with similar gene expression signature could also be similar in the global landscape of pathway activities. We also performed Gene Set Enrichment Analysis to identify pathways that associated with each subtype. As our particular interest in mesenchymal transition, published EMT signatures were also taken into analysis. At a FDR cutoff of 0.25, a total of 161 KEGG pathways were found to be significantly upregulated/downregulated in at least one subtype. 42 pathways were only dysregulated in one tissue and may represent tissue-specific processes. For example, 14 metabolism-associated pathways were found upregulated/ downregulated only in subtypes of hepatocellular carcinoma. On the other hand, 22 pathways were dysregulated in at least five tissues and therefore may represent common underlying mechanisms of carcinogenesis. ��Complement and coagulation cascades’ was the most frequently perturbed pathway, as it was dysregulated in AbMole Diperodon eleven subtypes. GSEA of EMT signatures was consistent with the mesenchymal phenotype of claudin-low, Mes and argued that Hoshida’s S1 and ovarian cancer C1 could also be mesenchymal subtypes. Interestingly, downregulated arm of two EMT signatures were found significantly downregulated in C5. DAVID functional analysis showed that all the EMT signatures we used did not overlap with any immune-associated pathway. When considering these five subtypes only, eleven pathways were downregulated only in C5 but upregulated in all other four mesenchymal subtypes. ��Complement and coagulation cascades’ was downregulated in both C5 and Hoshida’s S1, but upregulated in C1, Mes and claudin-low. In general, biological characteristics of molecular cancer subtypes could be defined by their gene expression signatures. For example, mesenchymal subtypes were usually defined by the overexpression of mesenchymal markers and underexpression of epithelial markers. Thus it may not be surprised to find common signature genes for cancer subtypes with similar biology. Instead of comparing signature genes, our study provided a systematic analysis focusing on genomewide transcriptional profile and pathway profile.