Fig. 5

Pan cancer analysis identified common metabolic pathways contributing to tumor fitness across all cancer types. Top panel: predicting differential expression across all cancers. Genes that impact flux through oxidative phosphorylation are more likely to be downregulated in tumors. Middle panel: predicting copy number variation across all cancers. The very low predictive accuracy (65%) suggests that most features predictive of CNV are not conserved between cancers. Bottom panel: predicting patient survival across all cancers. Genes that impact flux through pyruvate, glycine, serine, and threonine metabolism, glycolysis, oxidative phosphorylation, arginine and proline metabolism, and folate metabolism are predictive of patient survival across all cancers. Notably, expression of genes highly expressed in NCI-60 cancer cell lines is associated with poor survival