Fig. 4: Landscape of AID-related neoepitopes and its relation with ICI response.

a Percentage of neoepitopes originating from clonal/subclonal mutations in which the color indicates comparison for immunogenic or non-immunogenic calculated by Prime (top left; Two-sided two-sample Z-test for equal proportions). Top right shows the comparison of the percentage of samples having at least one AID immunogenic clonal neoepitopes (ICN) versus APOBEC ICN (“Presence”; Two-sided two-sample Z-test for equal proportions). Bottom plot shows the cumulative distribution of hotspot mutation utilization among the AID/APOBEC ICNs as a function of decreasing population-level frequency (Two-sided Mann–Whitney U test, FC of median AID to APOBEC values, error bars are 95% binomial confidence intervals). The number of distinct mutations indicates the number of unique samples having a mutation at a specific residue (e.g., IDH1 R132). b Heatmap of gene expression comparison between AID ICN “Presence” versus “Absence” groups across tumor types/all tumors (n = 2143; two-sided Wilcoxon test) measured as log2 FC. c OS prediction within Ipi-Naive patients improves when using AID ICN load (top right), ICN UV load (bottom right), ICN load (middle), or clonal neoepitopes load (left), lowest to highest log-rank p-values. d DEGs (p-adj < 0.20) between high ICN load patients versus low ICN load for pre-therapy, where increasing negative values on the x-axis shows higher significance (+Log10[p-adj]), or on-therapy, where increasing positive values on the x-axis means higher significance (−Log10[p-adj]). The y-axis shows upregulated (FC > 0) or downregulated genes (FC < 0) and colors indicate genes enriched in a specific pathway by GO analysis. a and b correspond to the TCGA cohort (n = 2143) meanwhile c and d to ICI-treated melanoma cohort (Riaz et al., n = 68).