The invention relates to a method for predicting a tumor newly-born
antigen. The method comprises the steps that 1, according to a tumor-embryonal
system contrast sample, somatic
mutation and
gene fusion detection are conducted; 2, for each pair of fusion genes, fusion
mutation peptide and corresponding wild
peptide are generated; 3, based on each somatic
mutation, mutation
peptide and corresponding wild peptide are generated; 4, a specific individual
genome of a
tumor sample is established, and mutation peptide containing multiple mutations is generated; 5, the true and false of mutation peptide of single-mutation and multi-mutation are judged; 6, mutation peptide completely identical to wild
protein in other position sequence is removed; 7, HLA molecular
subtyping detection is conducted,the appetency of newly-born peptide and
HLA molecules is predicted, and newly-born peptide with high appetency is used as a candidate tumor newly-born
antigen. The invention further provides a corresponding device and a computer storage medium. By adopting the method and device and the storage medium, the biomarker assessment can be effectively responded to through tumor treatment, and the precise candidate
peptide fragment is provided for design of a tumor vaccine.