Accounting for proximal variants improves neoantigen prediction.

Area of Science:

• Computational biology
• Cancer immunology
• Genomics

Background:

• Personalized cancer immunotherapies rely on predicting neoantigens.
• Current prediction methods often overlook proximal variants that affect peptide sequences and binding.
• This oversight can lead to inaccuracies in neoantigen identification.

Purpose of the Study:

• To investigate the impact of proximal somatic and germline alterations on neoantigenic peptide sequences.
• To assess how these proximal variants influence neoantigen binding predictions.
• To quantify the errors introduced by not considering proximal variants in neoantigen prediction.

Main Methods:

• Analysis of somatic variants from 430 tumor samples.
• Evaluation of in-phase missense proximal variants alongside other somatic variants.
• Estimation of false discovery and false negative rates for major histocompatibility complex class I neoantigen peptides (lengths 8-11) with and without proximal variant correction.

Main Results:

• An average of 241 missense somatic variants were analyzed per sample.
• Approximately 5% of somatic variants had in-phase missense proximal variants.
• Without proximal variant correction, estimated rates were 6.9% false discovery and 2.6% false negative for neoantigen peptides.

Conclusions:

• Proximal variants significantly alter neoantigen peptide sequences and binding predictions.
• Incorporating proximal variant correction is crucial for improving the accuracy of neoantigen prediction.
• Accurate neoantigen prediction is essential for the development of effective personalized cancer immunotherapies.