Trade-offs inside the black box of neoantigen prediction

Success of precision neoantigen-based immunotherapies hinges on the selection of immunogenic neoantigens, yet currently neither large-scale datasets nor streamlined methods are available to achieve this goal. Müller et al. present a large experimental dataset resource along with machine learning-based models to classify immunogenic neoantigens. Success of precision neoantigen-based immunotherapies hinges on the selection of immunogenic neoantigens, yet currently neither large-scale datasets nor streamlined methods are available to achieve this goal. Müller et al. present a large experimental dataset resource along with machine learning-based models to classify immunogenic neoantigens. Machine learning methods and harmonized datasets improve immunogenic neoantigen predictionMüller et al.ImmunityOctober 9, 2023In BriefMüller and colleagues showcase the enhanced capability of machine learning classifiers, which were trained using consistently processed multi-cancer genomic, transcriptomic, and neoantigen immunogenicity data. Their approach improves the prioritization of immunogenic neoantigens by incorporating additional features that complement factors related to antigen presentation and expression. Full-Text PDF Open Access