Fig. 5

GANB captures the immunogenomic characteristics of tumor cells. a Relationship between tumor neoantigen burden (y-axis) and tumor mutation burden (x-axis). The linear regression line is shown in the plot. b Proportions of TNB-low patients that can be re-defined by GANB as GANB-high across cancer types. c Heatmap of 1188 PCAWG tumors (columns) classified into four distinct TME subtypes based on unsupervised clustering of the 29 pre-defined gene signatures (rows). d Heatmap of the differences in TNB (left), GANB (middle), and SVNB (right) between immune-enriched tumors and immune-depleted tumors (two-sided Wilcoxon rank-sum test). Red indicates a higher neoantigen load in immune-enriched tumors while blue indicates a higher neoantigen load in immune-depleted tumors. *P < 0.05, **P < 0.005. e Differences in TNB/GANB between immune-enriched and immune-depleted tumors from Breast-AdenoCA (left) and Ovary-AdenoCA (right) (two-sided Wilcoxon rank-sum test). f Correlation between TIS and neoantigen load measured by TNB or GANB in Breast-AdenoCA and Ovary-AdenoCA (Spearman correlation). g Kaplan–Meier curves of patients with ovary adenocarcinoma stratified by TNB (left) and GANB (right), using median values as cutoffs (two-sided rank sum test)