Motivation: Splice variant neoantigens are a potential source of tumor-specific antigen (TSA) that are shared between patients in a variety of cancers, including acute myeloid leukemia. Current tools for genomic prediction of splice variant neoantigens demonstrate promise. However, many tools have not been well validated with simulated and/or wet lab approaches, with no studies published that have presented a targeted immunopeptidome mass spectrometry approach designed specifically for identification of predicted splice variant neoantigens.
Results: In this study, we describe NeoSplice, a novel computational method for splice variant neoantigen prediction based on (i) prediction of tumor-specific k-mers from RNA-seq data, (ii) alignment of differentially expressed k-mers to the splice graph and (iii) inference of the variant transcript with MHC binding prediction. NeoSplice demonstrates high sensitivity and precision (>80% on average across all splice variant classes) through in silico simulated RNA-seq data. Through mass spectrometry analysis of the immunopeptidome of the K562.A2 cell line compared against a synthetic peptide reference of predicted splice variant neoantigens, we validated 4 of 37 predicted antigens corresponding to 3 of 17 unique splice junctions. Lastly, we provide a comparison of NeoSplice against other splice variant prediction tools described in the literature. NeoSplice provides a well-validated platform for prediction of TSA vaccine targets for future cancer antigen vaccine studies to evaluate the clinical efficacy of splice variant neoantigens.
Availability and implementation: https://github.com/Benjamin-Vincent-Lab/NeoSplice.
Supplementary information: Supplementary data are available at Bioinformatics Advances online.
© The Author(s) 2022. Published by Oxford University Press.