Runaway mutations cause tumors, some of the non-synonymous mutational (i.e., missense, frameshift, fusion) peptides are degraded into short peptides by proteasome, then APCs (antigen-presenting cells) recognize and present them to draining lymph nodes. These short peptides, in line with the combination of MHC (major histocompatibility complex) motif, are captured by the T cell surface factors and produce an immune response, eventually lead to tumor regression. We call these peptides neoantigens, because these antigens are not negatively screened by thymus gland, they are recognized as "alien" by T cells and are not susceptible to immune tolerance mechanism. Neoantigens can act as effective targets for immune-mediated tumor control. The next generation sequencing technology greatly boosts the feasibility of neoantigen vaccine design, however identifying tumor somatic mutations that can be presented and recognized by TCR (T-cell receptor) is a very demanding yet key step. Many of the predicted "positive" neoantigen peptides are actually "false" and need to be eliminated in further steps to clinical application. Therefore effective screening method is an indispensable part of neoantigen vaccine therapy. However, no related reports have been seen in domestic. In this paper, the computational prediction process of screening methods for potential neoantigen peptides are reviewed.