Table S15 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data

Shapley values significantly associated with cancer types. Shapnorm = shapley effect score normalized across all predictions of the cancer type, stat_shap and pval_shap = bonferonni corrected outputs of Mann-Whitney U tests across distribution of Shapley value scores across cancer types vs non-cancer type predictions, shap_rank_ct = rank of the feature association within the cancer type it is associated with (i.e. 1 = top feature)