Fig. 2

Computational workflow of neoantigen prediction. a The general route of neoantigen prediction. b The workflow of clinical sample collection and single-cell sequencing. c The neoantigen sources can develop at the genomic level through SNV mutation, INDEL mutation, fusion mutation, integrated viral ORF and splice variants (the display of prediction software for mutation calling), at the transcriptomic level through alternative splicing, polyadenylation (pA), RNA editing and allegedly noncoding regions, and at the proteomic level through dysregulated translation and PTMs. d HLA typing prediction and display by prediction software tools. e and f Mutant polypeptides are produced by proteasome-mediated decomposition of endogenous proteins, which are subsequently transported to the ER by antigen-processing associated transporters (TAP). They may be loaded into MHC-I and MHC-II for binding to specific peptides produced by mutated proteins that breakdown in the endosomal pathway. These peptide-MHC-II/MHC-I (pMHC) complexes are then transported to the cell surface, where they are recognized by T cells. g pMHC complex binding prediction and the display of prediction software tools. h The prediction of T-cell recognition of pMHC complexes and the display of prediction software tools. i T-cell validation of neoantigens. Coculture of patient TILs or PBMCs with autologous antigen-presenting cells (APCs) expressing candidate neoantigens (TMG or peptides) allows for the identification of neoantigen-reactive T cells based on functional data such as IFN-γ release or 4-1BB expression. On the one hand, it could be injected into patients for cell therapy. On the other hand, the related functions of neoantigen reactive T cells have been verified by different experiments