This work evaluates the predictive strength of a composite biomarker, neoantigen presentation score (NEOPS), in a cohort of patients with late-stage melanoma. NEOPS incorporates damaging events in the antigen presentation machinery with predicted neoantigens to stratify patient response to immunotherapy. Others have previously reported that tumor mutational burden and RNA-based signatures are associated with therapeutic response, albeit with limited reproducibility. Data presented here demonstrate that our integrative approach outperformed such single-analyte biomarkers, suggesting that more complex models capturing multiple aspects of tumor escape may provide more robust stratification of patient response. We also demonstrate that data intensive biomarkers such as NEOPS are clinically practical, with comprehensive tumor profiling in our clinical cohort achieved using limited tumor tissue. These findings provide a novel composite biomarker of response in patients with late-stage melanoma, as well as evidence supporting the use of whole-exome and whole-transcriptome data in a clinical setting.