Extended Data Fig. 4: GIE association with cancer genomic features.

a) Heatmap displaying the association of genomic features with GIE frequency across cancer types. Significant associations that can not be explained by higher background mutation rate are highlighted by a red border line. Significant associations found in >2% of the GIE simulations are highlighted by a black border line. b) Comparison of the TMB between samples bearing GIE alterations and non-GIE samples in diffuse large B-cell lymphoma (DLBCL). Left boxplot, using observed GIE events. Right using simulated GIE events. c) Similar comparison for neoepitope burden in skin melanoma. d) Comparison of the APOBEC mutational exposure between samples bearing simulated GIE alterations and wild-type (no simulated GIE) in breast cancer in one randomly selected simulation. e) Analogous for ultraviolet light (UV) associated double base substitutions (DBSs) in skin melanoma and f) for platinum treatment associated DBSs in non-small cell lung cancer (NSCLC). g) Comparison of immune infiltration estimates from Davoli et al.³⁷ between samples bearing simulated GIE alterations and wild-type (no simulated GIE) in colorectal cancer. Boxplots: center line, median; First section out from the centerline contains 50% of the data. The next sections contain half the remaining data until we are at the outlier level. Each level out is shaded lighter. N, number of samples. P-values of the boxplots are calculated using a two-sided Mann–Whitney U test. One of the 100 simulations was randomly selected for all the simulated GIE boxplots. SBS, single base substitution.