Improved peptide-MHC class II interaction prediction through integration of eluted ligand and peptide affinity data
- 1 Downloads
Major histocompatibility complex (MHC) class II antigen presentation is a key component in eliciting a CD4+ T cell response. Precise prediction of peptide-MHC (pMHC) interactions has thus become a cornerstone in defining epitope candidates for rational vaccine design. Current pMHC prediction tools have, so far, primarily focused on inference from in vitro binding affinity. In the current study, we collate a large set of MHC class II eluted ligands generated by mass spectrometry to guide the prediction of MHC class II antigen presentation. We demonstrate that models developed on eluted ligands outperform those developed on pMHC binding affinity data. The predictive performance can be further enhanced by combining the eluted ligand and pMHC affinity data in a single prediction model. Furthermore, by including ligand data, the peptide length preference of MHC class II can be accurately learned by the prediction model. Finally, we demonstrate that our model significantly outperforms the current state-of-the-art prediction method, NetMHCIIpan, on an external dataset of eluted ligands and appears superior in identifying CD4+ T cell epitopes.
KeywordsMHC class II Ligand prediction CD4+ epitope Pan method Machine learning Mass spectrometry Peptidomics
ECJ was supported by a PhD stipend from the Innovation Fund Denmark. AWP was supported by a Principal Research Fellowship and project grant 1022509 from the National Health and Medical Research Council of Australia (NHMRC). SHR was supported by an Australian Postgraduate Award. MN is a researcher at the Argentinean National Research Council (CONICET).
We are grateful to the members of Prof. Purcell’s lab for their assistance in compiling the eluted ligand data. Furthermore, we thank the bioinformatics team at Evaxion Biotech for constructive feedback.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 6.Saini SK, Rekers N, Hadrup SR (2017) Novel tools to assist neoepitope targeting in personalized cancer immunotherapy. Ann Oncol 28:xii3–xii10Google Scholar
- 9.Andreatta M, Trolle T, Yan Z, Greenbaum JA, Peters B et al (2018) An automated benchmarking platform for MHC class II binding prediction methods. Bioinformatics 34:1522–1528Google Scholar
- 10.Fleri W, Paul S, Dhanda SK, Mahajan S, Xu X et al. (2017) The immune epitope database and analysis resource in epitope discovery and synthetic vaccine design. Front Immunol; 8:278Google Scholar
- 13.EMBL-EBI (2018) IPD-IMGT/HLA database—statistics. Available at: https://www.ebi.ac.uk/ipd/imgt/hla/stats.html [Accessed July 4, 2018]
- 18.Shilov IV, Seymour SL, Patel AA, Loboda A, Tang WH, Keating SP, Hunter CL, Nuwaysir LM, Schaeffer DA (2007) The paragon algorithm, a next generation search engine that uses sequence temperature values and feature probabilities to identify peptides from tandem mass spectra. Mol Cell Proteomics 6:1638–1655CrossRefGoogle Scholar
- 27.Bentzen AK, Marquard AM, Lyngaa R, Saini SK, Ramskov S, Donia M, Such L, Furness AJS, McGranahan N, Rosenthal R, Straten P, Szallasi Z, Svane IM, Swanton C, Quezada SA, Jakobsen SN, Eklund AC, Hadrup SR (2016) Large-scale detection of antigen-specific T cells using peptide-MHC-I multimers labeled with DNA barcodes. Nat Biotechnol 34:1037–1045CrossRefGoogle Scholar