A machine learning model that identifies neoantigen-reactive CD8+ T cells in human gastrointestinal cancer

It appears that tumor-infiltrating neoantigen-reactive CD8 + T cells are the primary driver of immune responses to gastrointestinal cancer in patients. By mapping neoantigen-reactive T cells from the single-cell transcriptomes of thousands of tumor-infiltrating lymphocytes, we developed a 26-gene machine learning model for the identification of neoantigen-reactive T cells. In both training and test sets, the model performed admirably. We discovered, by applying the model to large-scale single-cell sequencing data of tumor-infiltrating CD8 + T cells, that Neo T cells exhibited a hyperexpanded phenotype and two distinct differentiation pathways. Moreover, compared to non-neoantigen-reactive T cells, the majority of neoantigen-reactive T cells exhibited notable differences in the biological processes of locomotion and amide metabolism. The analysis of potential cell-to-cell interactions revealed that neoantigen-reactive T cells contain potent signaling molecules, such as CXCL13 and LTA, associated with the formation of tertiary lymphoid structures. This method expedites the identification of neoantigen-reactive TCRs and the engineering of neoantigen-reactive T cells for therapy.