Abbott2021 - Prediction of Immunotherapy Response in Melanoma through Combined Modeling of NB and Immune-Related Resistance Mechanisms

In this manuscript, the authors had hypothesized a multi-dimensional approach modeling of both tumor and immune-related molecular mechanisms would better predict immune checkpoint blockade (ICB) response than simpler mutation-focused biomarkers, such as tumor mutational burden (TMB). The authors showed that the predictive power increases with deeper modeling of neoantigens and immune-related resistance mechanisms of ICB. The neoantigen burden score (NBS) and composite neoantigen presentation score (NEOPS) mentioned in the transcript was fully reproduced. Internally they used XGBoost algorithm to generate the results and the same is provided as dataset file. That is, the dataset provided here demonstrates that their integrative approach outperformed single-analyte biomarkers such as those found in cohort of patients with late-stage melanoma. This model is now addresses the issues in reproducing itself which was caused by version changes and deprecation of some R packages. It uses checkpoint package, which acts as a time machine for CRAN packages thereby promoting FAIReR sharing of ML models.