Fig. 3

The workflow of computational prediction and screening pipelines for neoantigens To identify tumor-specific somatic mutations, tumor tissue and normal tissue samples (usually peripheral blood mononuclear cells) are acquired from the patient perform WES/WGS. Additional RNA sequencing provides information on the gene expression of the mutated genes and further confirmation of gene fusion. Peripheral blood cells were used to predict HLA typing performed by RNA-seq or DNA-seq analysis. MHC-peptide binding prediction software predicts peptides presented by MHC. Computational filtering/screening involves three levels: filter 1 is based on RNA expression; filter 2 is based on proteomics mass spectrometry identification; filter 3 is based on database high confidence filtering. By integrating various physical and chemical properties of peptides, computational prediction screening also includes three levels: antigen binding; neoantigen peptide epitope-TCR recognition; immunogenicity calculation to prioritize the predicted peptides and screen out the neoantigens with high confidence that could be recognized by TCR