Fig. 5

A Overlap of TCGA samples between all data types used in mutation prediction comparisons. Only overlaps with more than 100 samples are shown. Somatic mutation sample information is included because it is needed to generate the mutation presence/absence labels. B Overall distribution of performance per data type across 217 genes from the cancer-related gene set. Each data point represents mean cross-validated AUPR difference, compared with a baseline model trained on permuted labels, for one gene; notches show bootstrapped 95% confidence intervals. Significance stars indicate results of Bonferroni-corrected pairwise Wilcoxon tests: **p < 0.01, ***p < 0.001, ns: not statistically significant for a cutoff of p = 0.05. All pairwise tests were run, and corrected for, but only neighboring test results are shown. C Overall performance distribution per data type for genes where the permuted baseline model is significantly outperformed for one or more data types, resulting in a total of 39 genes. D–F Volcano-like plots showing predictive performance for each gene in the cancer-related gene set, in each of the added data types (RPPA, microRNA, mutational signatures). The x-axis shows the difference in mean AUPR compared with a baseline model trained on permuted labels, and the y-axis shows p-values for a paired t-test comparing cross-validated AUPR values within folds. G–I Direct comparison of performance using gene expression and each added data type, showing only genes that perform significantly better than the baseline model for both data types. Points (genes) to the left of y=0 perform better using gene expression-derived features, and points to the right perform better using the added data type (RPPA, microRNA, and mutational signatures respectively). J Pan-cancer survival prediction performance, quantified using c-index on the y-axis, for all data types. The x-axis shows results with varying numbers of principal components included for each data type. Models also included covariates for patient age, sample mutation burden, and sample cancer type; gray dotted line indicates mean performance for a covariate-only baseline model