[Abstract] Objective: To screen candidate epitopes of breast cancer HLA-A2 restrictive neoantigen and to identify high frequency mutation sites in breast cancer neoantigen by using bioinformatics method. Methods: NCBI and GDC databases were used to search missense mutation sites formed by single nucleotide mutation in breast cancer among reported literatures and sequencing data. The new antigen epitopes were predicted by HLA-A2 antigen epitope prediction website BIMAS, SYFPEITHI and artificial neural networkbased NetMHC4.0, and the epitopes with TAP binding power less than Intermediate were eliminated. The candidate epitopes were prioritized by mutation frequency and prediction results. Results: A total of 17 high-frequency mutation genes, including BTLA, ERBB2 and NBPF12 etc, were screened by the above-mentioned methods, and a total of 26 neoantigen epitopes were identified. The binding power of epitopes predicted using BIMAS and SYFPEITHI showed great difference (P<0.05), epitopes in high priority as GSTP1 (A114V , mutation frequency of 5.94%) and BRCA2 (N991H, mutation frequency of 5.40%) etc, were expected to be candidate neo-antigen epitopes; however, their mutation frequency was relatively too low to achieve“universal use” . The possibility of these epitopes used as general breast cancer neo-antigen epitopes is less likely. Conclusion: The common mutation frequency of breast cancer is lower than that of other tumors; it ’s difficult to find“universal”new antigen epitopes of breast cancer; the individualized neoantigen vaccine may be of more promise, which needs further research.