Fig. 3: Mutated FABP-derived neoantigens are enriched in primary tumor lines from DNGR-1-deficient mice.

a, Histogram showing the number of predicted neoantigens according to binding affinity (IC₅₀) scores, stratified according to progressor and regressor cell lines. b, Box and whisker plots comparing the total number of predicted strong neoantigens (binding affinity IC₅₀ < 150 nM) between progressor (n = 22) and regressor (n = 21) cell lines (two-sided Mann-Whitney U-test P = 0.2). Box plots display the median (central line) and the first and third quartiles (lower and upper box edges). Whiskers extend to 1.5× the interquartile range from each quartile. c–e, As in b, box and whisker plots comparing the number of predicted strong neoantigens (binding affinity IC₅₀ < 150 nM) between cell lines from different host background (WT, DNGR-1^KO, RAG1^KO with n = 18, n = 14 and n = 11, respectively) in all (WT versus DNGR-1^KO, P = 0.75; WT versus RAG1^KO, P = 0.22; WT versus RAG1^KO, P = 0.34) (c), FABP (WT versus DNGR-1^KO, P = 0.049; WT versus RAG1^KO, P = 0.034; WT versus RAG1^KO, P = 0.84) (d), and MBP genes (WT versus DNGR-1^KO, P = 0.37; WT versus RAG1^KO, P = 0.42; WT versus RAG1^KO, P = 0.25) (e). The only significant differences are between the number of predicted FABP neoantigens in DNGR-1^KO or RAG1^KO lines versus WT. f, Heatmap showing strong neoantigens predicted within the FABP genes in immunogenic regressor cell lines, clustered by primary host background of origin.