How Scientists Use Computers to Design Personalized Peptide Cancer Vaccines

The neoantigen prediction workflow involves multiple computational steps: somatic mutation identification from tumor-normal sequencing, HLA typing to determine the patient's immune molecule profile, peptide processing prediction, and peptide-MHC binding prediction. The authors provide specific recommendations for each step and identify key areas needing improvement, including HLA class II typing accuracy, software support for diverse neoantigen sources beyond point mutations, and incorporation of clinical response data to improve prediction algorithms. Currently, there is no consensus approach, and the field needs standardization.

This is a comprehensive review and best-practices guide that evaluates computational tools and analysis workflows for neoantigen characterization. The authors reviewed the literature on neoantigen prediction, prioritization, delivery, and validation, synthesizing findings from multiple preclinical and clinical trials to provide practical guidance for clinical implementation.

Personalized cancer vaccines are one of the most promising frontiers in oncology. The success of these vaccines depends entirely on correctly identifying which peptide neoantigens to include. This guide helps researchers and clinicians avoid common pitfalls and apply best practices, potentially improving vaccine efficacy and patient outcomes.

Personalized neoantigen vaccines have shown dramatic results in clinical trials, including long-lasting remissions in melanoma and other cancers. But the technology is only as good as the computational predictions that identify which peptides to use. This review addresses a critical bottleneck in the field — standardizing how we select the right neoantigens to target.

As a review and best-practices guide, no new experimental data is presented. The field is evolving rapidly, and some recommendations may be superseded by newer tools and algorithms. Prediction accuracy for neoantigen immunogenicity remains imperfect — not all predicted neoantigens actually trigger immune responses. HLA class II prediction is notably less accurate than class I.