Tumor neoantigens are important targets for immunotherapy. Based on high-throughput tumor genomic analysis, each missense mutation can potentially give rise to multiple neopeptides, numerous false positive neoantigens will be produced, experimental verification would require immense time and experience. Specific identification of immunogenic candidate neoantigens is consequently a major challenge. Here we introduce a workflow to predict and filter neoantigens of breast cancer with proteogenomic methodology, which can be beneficial to high quality identification of neoantigens. We found that C2 (IFN-γdominant) immunophenotype possessed the most number of neoantigens. C2 immunophenotype shows the highest M1/M2 macrophage polarization, a strong CD8 signal and the greatest T cell receptor(TCR) diversity, which may lead to more neoantigen number than in other immunophenotypes of breast cancer. In addition, we also observed that there is a positive linear relationship between neoantigen number and tumor mutation burden. By further screening for predicted tumor mutant peptides using mass spectral data of breast cancer, we found that more than 20 000 predicted neoantigens were reduced to dozens of mutant peptides in protein expression level, the corresponding mutant genes could be further analyzed. Finally, in order to define which neoantigens were more likely to be immunogenic, TCR recognition probability was calculated using blastp method. In this study, proteomics data was used to further screen the predicted neoantigens, which improved the prediction accuracy of neoantigens, and could greatly reduce the validation scope of potential subsequent experiments. This workflow provides a new insight for tumor neoantigen prediction and screening.