This review provides best practices for using bioinformatics to identify and prioritize neoantigen peptides for personalized cancer vaccine development.
No consensus approach existsDespite multiple clinical trials using neoantigen vaccines, there are still no standardized best practices for the computational prediction pipeline
What the researchers found
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.
Why it matters
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.
How the study worked
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.
What this study cannot tell us
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 The field is evolving rapidly, and some recommendations may be superseded by newer tools and algorithms.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.
How to read the evidence
This is a comprehensive review and best-practices guide synthesizing evidence from multiple preclinical and clinical studies. It provides expert recommendations but no new experimental data.
When this study was published
Published in 2019, this review captures the state of neoantigen bioinformatics during a period of rapid advancement. Some tools and algorithms have improved since publication, but the fundamental workflow and challenges described remain relevant.
The bigger picture
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.
Questions still open
- How can clinical response data be systematically incorporated to improve neoantigen prediction algorithms?
- Will improved HLA class II typing accuracy significantly increase vaccine efficacy by enabling CD4+ T cell responses?
- Can AI and machine learning overcome the current limitations in predicting which neoantigens will actually be immunogenic?
Common questions
What are neoantigens and how are they used in cancer vaccines?
Why is the computational prediction step so important?
Read the original research
Best practices for bioinformatic characterization of neoantigens for clinical utility.
Genome medicine, 11(1), 56
Citation
Richters, Megan M; Xia, Huiming; Campbell, Katie M; Gillanders, William E; Griffith, Obi L; Griffith, Malachi. (2019). Best practices for bioinformatic characterization of neoantigens for clinical utility.. Genome medicine, 11(1), 56. https://doi.org/10.1186/s13073-019-0666-2