Skip to main content
Fig. 6 | BMC Cancer

Fig. 6

From: Population-level distribution and putative immunogenicity of cancer neoepitopes

Fig. 6

Neoepitope prioritization metric scores and linear modeling in the cohort of peptides with neoepitope-specific immune response data. Score distributions are shown for peptides with (red) and without (gray) neoepitope-specific immune response. Panes in light blue indicate metrics which are included in the model shown in part G. R = “response”, NR = “nonresponse”. a. Neoepitope binding affinity, * p < 0.001 in a Wilcoxon rank sum test. b. Difference in binding affinity between the neoepitope and its paired normal epitope (normal affinity – tumor affinity). c. Percent protein sequence similarity (see Methods) between the neoepitope and its paired normal epitope. d. Percent protein sequence similarity between the neoepitope and its closest human peptide. e. Percent protein sequence similarity between the neoepitope and its closest bacterial peptide. f. Percent protein sequence similarity between the neoepitope and its closest viral peptide. g. ROC curve for prediction of immunogenicity from prioritization criteria. A linear model incorporating neoepitope binding affinity, protein sequence similarity between neoepitopes and their closest viral peptides, and difference in binding affinity between neoepitopes and their closest human peptides was used to predict immune response. Within a limited cohort of 419 peptides with immune response data, our model was able to predict peptide immunogenicity with an AUROC of 0.66. Dashed gray line represents the line y = x for comparison

Back to article page