Fig. 1

Computational workflow for the prediction of metabolite-gene-pathway sets (MGPs). Computational workflow for predicting MGPs using cancer patient-specific genome-scale metabolic models (GEMs). This workflow is repeated for each metabolite against a list of mutated genes in cancers. The computational workflow requires RNA-seq data and mutation data for each cancer sample. Flux-sum value for a target metabolite is first predicted using a cancer patient-specific GEM that is generated using RNA-seq data (step 1). Next, a metabolite is paired with a gene if flux-sum distributions of the metabolite appear to be significantly different upon mutation of the gene (step 2). Metabolite-gene (MG) pairs predicted from the previous step are connected with metabolic pathways that biosynthesize a target metabolite if these pathways show significantly different “target flux-sum” values upon mutation of a target gene (step 3). MG pairs from the previous step are removed if such target pathways are not found. Finally, MGPs are selected by identifying target genes in each target pathway that show target flux-sum values significantly different from those of other target genes in the same pathway (step 4). For this, for each target gene in a target pathway, the mean of its target flux-sum values is calculated, and converted to the modified Z-score. The selected MGPs should have their modified Z-score satisfying the threshold of “3.5”