Fig.6.

Prognostic modules associated with outcome in bladder cancer patients with high-grade, muscle invasive tumors. In each cohort, (A) over-represented Gene Ontology (GO) terms and (B) KEGG pathways were identified from lists of genes significantly predictive of disease outcome (P <  0.01) using the DAVID gene annotation enrichment analysis toolkit. Consistently prognostic modules were identified by ranking all modules first by the number of cohorts with significant results (FDR <  20% ) and then by average p-value. Each figure includes ten modules: the most consistently prognostic modules and the ‘top hit’ for each cohort, marked by an asterisk (*), which is defined as the module with the lowest FDR in that cohort that has an FDR <  20% in multiple cohorts, or if no such module exists, then the module with the lowest FDR.