Artificial intelligence: accelerating translational cell therapy development through computational methods for neoepitope discovery

Abstract

A surge in the efforts to identify cancer-specific antigens as targets for immunotherapeutic approaches has characterized the past few decades. However, the clinical use of tumor-associated antigens has been mostly limited to cancer/testis antigens, restricting its application to certain cancers and limiting the weak immune response elicited by self-antigens. Neoantigens resulting from somatic mutations in cancer cells represent an alternative immunotherapy target voided of such limitations. Similarly, cancer-specific protein isoforms or mutational hotspot antigens provide the advantage of tumor tissue specificity coupled with low thymic selection and central tolerance. Choosing the best antigenic determinants is critical for successful adoptive cell therapy; the availability of algorithms for clinical data mining and to predict the most immunogenic epitopes is thus a powerful promise for the rapid advancement of cancer medicine. The state of the art of computational neo-antigen research is discussed here.