A novel computational approach to predict metabolites associated with somatic mutations in cancers

Abstract

A representative byproduct of metabolic reprogramming in cancers is oncometabolites, which are defined to be metabolites that show abnormal accumulation in a cancer cell, and are generated upon mutations in a metabolic gene, for example, IDH, SDH and/or FH mutations in gliomas and acute myeloid leukemia (AML). Also, even if a metabolite is not an oncometabolite, it may still be associated with a specific mutation, for example, mevalonate pathway intermediates associated with TP53 mutation. Here, identifying additional metabolite-mutation associations can facilitate establishing novel drug targets and biomarkers for cancer treatment and diagnosis. In this study, a novel computational workflow was developed that predicts metabolites and relevant pathways associated with gene mutation for 24 cancer types by using cancer patient-specific genome-scale metabolic models and mutation. This computational workflow was validated using multi-omics data of AML samples collected in this study. Our computational approach and its prediction outcomes can help better understand the mutationassociated metabolic reprogramming in cancers, and be valuable resources for future studies on cancer metabolism.