Additional file 6 of Systematic interrogation of mutation groupings reveals divergent downstream expression programs within key cancer genes

Additional file 6: Figure S5. Subgrouping prediction tasks outperform gene-specific random background prediction tasks. In addition to our cohort-specific null background tasks, we also constructed gene-specific random tasks for each tested subgrouping by randomly selecting five size-matched groups from the set of samples carrying any point mutation of the gene in the cohort. The AUCs of these gene-specific random tasks (grey distributions) tended to be lower than the AUCs of subgroupings (colorful distributions) in genes that had at least one subgrouping task with an AUC of 0.7. Distributions of subgroupings containing at least one subgrouping with performance cv-significantly better than that of the gene-wide task (dashed line) are annotated with an asterisk; genes in which at least one random subgrouping was cv-significantly better than the best original subgrouping are also annotated with an asterisk. We include here these figures for the METABRIC-(LumA) and TCGA-BRCA(LumA) cohorts and for the remaining cohorts at our data portal under Figures/S5 - Gene Null Background AUC Comparisons. The names of these figures have the format (expr-source)__(cohort)__gene-comparisons_Ridge.svg.