NNAlign_MA; MHC Peptidome Deconvolution for Accurate MHC Binding Motif Characterization and Improved T-cell Epitope Predictions

The set of peptides presented on a cell's surface by MHC molecules is known as the immunopeptidome. Current mass spectrometry technologies allow for identification of large peptidomes, and studies have proven these data to be a rich source of information for learning the rules of MHC-mediated antigen presentation. Immunopeptidomes are usually poly-specific, containing multiple sequence motifs matching the MHC molecules expressed in the system under investigation. Motif deconvolution -the process of associating each ligand to its presenting MHC molecule(s)- is therefore a critical and challenging step in the analysis of MS-eluted MHC ligand data. Here, we describe NNAlign_MA, a computational method designed to address this challenge and fully benefit from large, poly-specific data sets of MS-eluted ligands. NNAlign_MA simultaneously performs the tasks of (1) clustering peptides into individual specificities; (2) automatic annotation of each cluster to an MHC molecule; and (3) training of a prediction model covering all MHCs present in the training set. NNAlign_MA was benchmarked on large and diverse data sets, covering class I and class II data. In all cases, the method was demonstrated to outperform state-of-the-art methods, effectively expanding the coverage of alleles for which accurate predictions can be made, resulting in improved identification of both eluted ligands and T-cell epitopes. Given its high flexibility and ease of use, we expect NNAlign_MA to serve as an effective tool to increase our understanding of the rules of MHC antigen presentation and guide the development of novel T-cell-based therapeutics. The set of peptides presented on a cell's surface by MHC molecules is known as the immunopeptidome. Current mass spectrometry technologies allow for identification of large peptidomes, and studies have proven these data to be a rich source of information for learning the rules of MHC-mediated antigen presentation. Immunopeptidomes are usually poly-specific, containing multiple sequence motifs matching the MHC molecules expressed in the system under investigation. Motif deconvolution -the process of associating each ligand to its presenting MHC molecule(s)- is therefore a critical and challenging step in the analysis of MS-eluted MHC ligand data. Here, we describe NNAlign_MA, a computational method designed to address this challenge and fully benefit from large, poly-specific data sets of MS-eluted ligands. NNAlign_MA simultaneously performs the tasks of (1) clustering peptides into individual specificities; (2) automatic annotation of each cluster to an MHC molecule; and (3) training of a prediction model covering all MHCs present in the training set. NNAlign_MA was benchmarked on large and diverse data sets, covering class I and class II data. In all cases, the method was demonstrated to outperform state-of-the-art methods, effectively expanding the coverage of alleles for which accurate predictions can be made, resulting in improved identification of both eluted ligands and T-cell epitopes. Given its high flexibility and ease of use, we expect NNAlign_MA to serve as an effective tool to increase our understanding of the rules of MHC antigen presentation and guide the development of novel T-cell-based therapeutics. Major Histocompatibility Complex (MHC)1 molecules play a central role in the cellular immune system as cell-surface presenters of antigenic peptides to T-cell receptors (TCR). On presentation, the peptide-MHC complex (pMHC) is scrutinized by T cells and an immune response can be initiated if interactions between the pMHC and TCR are established. The collection of peptides presented by MHC molecules is referred to as the immunopeptidome. Because of the extreme polymorphism of the MHC, immunopeptidomes can vary dramatically within a population, contributing to the personalized attributes of the vertebrate immune system. Because of the essential role of the MHC in defining immune responses, large efforts have been dedicated to understanding the rules that shape the immunopeptidome, as well as its alterations in disease–either as a result of pathogen infection or cancerous mutation (1.Caron E. Aebersold R. Banaei-Esfahani A. Chong C. Bassani-Sternberg M. A Case for a Human Immuno-Peptidome Project Consortium.Immunity. 2017; 47: 203-208Abstract Full Text Full Text PDF PubMed Scopus (49) Google Scholar). A crucial step toward defining the immunopeptidome of an individual is the characterization of the binding preferences of MHC molecules. The peptide-binding domain of MHC molecules consists of a groove, with specific amino acid preferences at different positions. MHC class I, by and large, loads peptides between eight and thirteen residues long (2.Trolle T. McMurtrey C.P. Sidney J. Bardet W. Osborn S.C. Kaever T. Sette A. Hildebrand W.H. Nielsen M. Peters B. The length distribution of Class I-restricted T cell epitopes is determined by both peptide supply and MHC allele-specific binding preference.J. Immunol. 2016; 196: 1480-1487Crossref PubMed Scopus (103) Google Scholar, 3.Gfeller D. Guillaume P. Michaux J. Pak H.-S. Daniel R.T. Racle J. Coukos G. Bassani-Sternberg M. The length distribution and multiple specificity of naturally presented HLA-I ligands.J. Immunol. 2018; 201: 3705-3716Crossref PubMed Scopus (64) Google Scholar). MHC class II molecules have an open binding groove at both ends and can bind much longer peptides, and even whole proteins (4.Sette A. Adorini L. Colon S.M. Buus S. Grey H.M. Capacity of intact proteins to bind to MHC class II molecules.J. Immunol. 1989; 143: 1265-1267PubMed Google Scholar, 5.Mommen G.P.M. Marino F. Meiring H.D. Poelen M.C.M. van Gaans-van den Brink J.A.M. Mohammed S. Heck A.J.R. van Els C.A.C.M. Sampling from the proteome to the human leukocyte antigen-DR (HLA-DR) ligandome proceeds via high specificity.Mol. Cell. Proteomics. 2016; 15: 1412-1423Abstract Full Text Full Text PDF PubMed Scopus (49) Google Scholar). Peptide-MHC binding affinity (BA) assays represented the first attempts of studying binding preferences of different MHC molecules in vitro (6.Buus S. Sette A. Colon S.M. Miles C. Grey H.M. The relation between major histocompatibility complex (MHC) restriction and the capacity of Ia to bind immunogenic peptides.Science. 1987; 235: 1353-1358Crossref PubMed Scopus (544) Google Scholar, 7.Townsend A. Elliott T. Cerundolo V. Foster L. Barber B. Tse A. Assembly of MHC class I molecules analyzed in vitro.Cell. 1990; 62: 285-295Abstract Full Text PDF PubMed Scopus (411) Google Scholar). However, BA characterization ignores many in vivo antigen processing and presentation features, such as protein internalization, protease digestion, peptide transport, peptide trimming, and the role of different chaperones involved in the folding of the pMHC complex (8.Sadegh-Nasseri S. Kim A. MHC class II auto-antigen presentation is unconventional.Front. Immunol. 2015; 6: 372Crossref PubMed Scopus (27) Google Scholar). Further, BA assays most often are conducted one peptide at a time, thus becoming costly, time-consuming, and low-throughput. Recently, advances in liquid chromatography mass spectrometry (in short, LC-MS/MS) technologies have opened a new chapter in immunopeptidomics. Several thousands of MHC-associated eluted ligands (in short, EL) can with this technique be sequenced in a single experiment (9.Caron E. Kowalewski D.J. Chiek Koh C. Sturm T. Schuster H. Aebersold R. Analysis of major histocompatibility complex (MHC) immunopeptidomes using mass spectrometry.Mol. Cell. Proteomics. 2015; 14: 3105-3117Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar) and numerous assessments have proven MS EL data to be a rich source of information for both rational identification of T-cell epitopes (10.Abelin J.G. Keskin D.B. Sarkizova S. Hartigan C.R. Zhang W. Sidney J. Stevens J. Lane W. Zhang G.L. Eisenhaure T.M. Clauser K.R. Hacohen N. Rooney M.S. Carr S.A. Wu C.J. Mass spectrometry profiling of HLA-associated peptidomes in mono-allelic cells enables more accurate epitope prediction.Immunity. 2017; 46: 315-326Abstract Full Text Full Text PDF PubMed Scopus (341) Google Scholar, 11.Graham D.B. Luo C. O'Connell D.J. Lefkovith A. Brown E.M. Yassour M. Varma M. Abelin J.G. Conway K.L. Jasso G.J. Matar C.G. Carr S.A. Xavier R.J. Antigen discovery and specification of immunodominance hierarchies for MHCII-restricted epitopes.Nat. Med. 2018; 24: 1762Crossref PubMed Scopus (38) Google Scholar) and learning the rules of MHC antigen presentation (12.Bassani-Sternberg M. Pletscher-Frankild S. Jensen L.J. Mann M. Mass spectrometry of human leukocyte antigen class I peptidomes reveals strong effects of protein abundance and turnover on antigen presentation.Mol. Cell. Proteomics. 2015; 14: 658-673Abstract Full Text Full Text PDF PubMed Scopus (280) Google Scholar, 13.Barra C. Alvarez B. Paul S. Sette A. Peters B. Andreatta M. Buus S. Nielsen M. Footprints of antigen processing boost MHC class II natural ligand predictions.Genome Med. 2018; 10: 84Crossref PubMed Scopus (45) Google Scholar). In this context, we have demonstrated how a modeling framework that integrates both BA and EL data achieves superior predictive performance for T-cell epitope discovery compared with models trained on either of the two data types alone (13.Barra C. Alvarez B. Paul S. Sette A. Peters B. Andreatta M. Buus S. Nielsen M. Footprints of antigen processing boost MHC class II natural ligand predictions.Genome Med. 2018; 10: 84Crossref PubMed Scopus (45) Google Scholar, 14.Jurtz V. Paul S. Andreatta M. Marcatili P. Peters B. Nielsen M. NetMHCpan-4.0: improved peptide-MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data.J. Immunol. 2017; 199: 3360-3368Crossref PubMed Scopus (672) Google Scholar). In these studies, the modeling framework was an improved version of the NNAlign method (15.Nielsen M. Andreatta M. NNAlign: a platform to construct and evaluate artificial neural network models of receptor-ligand interactions.Nucleic Acids Res. 2017; 45: W344-W349Crossref PubMed Scopus (33) Google Scholar), which incorporated two output neurons to enable training and prediction on both BA and EL data types. In this setup, weight-sharing allows information to be transferred between the two data types resulting in a boost in predictive power. For MHC class I, we have demonstrated how this framework can be extended to a pan-specific model, capturing the specific antigen presentation rules for any MHC molecule with known protein sequence, including molecules characterized by limited, or even no, binding data (14.Jurtz V. Paul S. Andreatta M. Marcatili P. Peters B. Nielsen M. NetMHCpan-4.0: improved peptide-MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data.J. Immunol. 2017; 199: 3360-3368Crossref PubMed Scopus (672) Google Scholar, 16.Nielsen M. Connelley T. Ternette N. Improved prediction of bovine leucocyte antigens (BoLA) presented ligands by use of mass-spectrometry-determined ligand and in vitro binding data.J. Proteome Res. 2018; 17: 559-567Crossref PubMed Scopus (26) Google Scholar, 17.DeVette C.I. Andreatta M. Bardet W. Cate S.J. Jurtz V.I. Jackson K.W. Welm A.L. Nielsen M. Hildebrand W.H. NetH2pan: A computational tool to guide MHC peptide prediction on murine tumors.Cancer Immunol. Res. 2018; 6: 636-644Crossref PubMed Scopus (10) Google Scholar). Except genetically engineered cells, all nucleated cells express multiple MHC-I alleles and all antigen presenting cells additionally express multiple MHC-II alleles on their surface. The antibodies used to purify peptide-MHC complexes in MS EL experiments are mostly pan- or locus-specific, and the data generated in an MS experiment are thus inherently poly-specific - i.e. they contain ligands matching multiple binding motifs. For instance, in the context of the human immune system, each cell can express up to six different MHC class I molecules, and the immunopeptidome obtained using MS techniques will thus be a mixture of all ligands presented by these MHCs (12.Bassani-Sternberg M. Pletscher-Frankild S. Jensen L.J. Mann M. Mass spectrometry of human leukocyte antigen class I peptidomes reveals strong effects of protein abundance and turnover on antigen presentation.Mol. Cell. Proteomics. 2015; 14: 658-673Abstract Full Text Full Text PDF PubMed Scopus (280) Google Scholar). The poly-specific nature of MS EL libraries constitutes a challenge in terms of data analysis and interpretation, where, to learn specific MHC rules for antigen presentation, one must first associate each ligand to its presenting MHC molecule(s) within the haplotype of the cell line. Several approaches have been suggested to address this task, including experimental setups that employ cell lines expressing only one specific MHC molecule (10.Abelin J.G. Keskin D.B. Sarkizova S. Hartigan C.R. Zhang W. Sidney J. Stevens J. Lane W. Zhang G.L. Eisenhaure T.M. Clauser K.R. Hacohen N. Rooney M.S. Carr S.A. Wu C.J. Mass spectrometry profiling of HLA-associated peptidomes in mono-allelic cells enables more accurate epitope prediction.Immunity. 2017; 46: 315-326Abstract Full Text Full Text PDF PubMed Scopus (341) Google Scholar, 18.Prilliman K. Lindsey M. Zuo Y. Jackson K.W. Zhang Y. Hildebrand W. Large-scale production of class I bound peptides: assigning a signature to HLA-B*1501.Immunogenetics. 1997; 45: 379-385Crossref PubMed Scopus (62) Google Scholar, 19.Falk K. Rötzschke O. Stevanović S. Jung G. Rammensee H.G. Allele-specific motifs revealed by sequencing of self-peptides eluted from MHC molecules.Nature. 1991; 351: 290-296Crossref PubMed Scopus (2115) Google Scholar, 20.Schittenhelm R.B. Dudek N.L. Croft N.P. Ramarathinam S.H. Purcell A.W. A comprehensive analysis of constitutive naturally processed and presented HLA-C*04:01 (Cw4)-specific peptides.Tissue Antigens. 2014; 83: 174-179Crossref PubMed Scopus (37) Google Scholar), and approaches inferring MHC associations using prior knowledge of MHC specificities (21.Murphy J.P. Konda P. Kowalewski D.J. Schuster H. Clements D. Kim Y. Cohen A.M. Sharif T. Nielsen M. Stevanovic S. Lee P.W. Gujar S. MHC-I ligand discovery using targeted database searches of mass spectrometry data: implications for T-cell immunotherapies.J. Proteome Res. 2017; 16: 1806-1816Crossref PubMed Scopus (40) Google Scholar) or by means of unsupervised sequence clustering (22.Bassani-Sternberg M. Gfeller D. Unsupervised HLA peptidome deconvolution improves ligand prediction accuracy and predicts cooperative effects in peptide-HLA interactions.J. Immunol. 2016; 197: 2492-2499Crossref PubMed Scopus (72) Google Scholar). For instance, GibbsCluster (23.Andreatta M. Lund O. Nielsen M. Simultaneous alignment and clustering of peptide data using a Gibbs sampling approach.Bioinformatics. 2013; 29: 8-14Crossref PubMed Scopus (77) Google Scholar, 24.Andreatta M. Alvarez B. Nielsen M. GibbsCluster: unsupervised clustering and alignment of peptide sequences.Nucleic Acids Res. 2017; 45: W458-W463Crossref PubMed Scopus (91) Google Scholar) has been successfully employed in multiple studies to extract binding motifs from EL data sets of several species, both for MHC class I and MHC class II (5.Mommen G.P.M. Marino F. Meiring H.D. Poelen M.C.M. van Gaans-van den Brink J.A.M. Mohammed S. Heck A.J.R. van Els C.A.C.M. Sampling from the proteome to the human leukocyte antigen-DR (HLA-DR) ligandome proceeds via high specificity.Mol. Cell. Proteomics. 2016; 15: 1412-1423Abstract Full Text Full Text PDF PubMed Scopus (49) Google Scholar, 25.Ritz D. Gloger A. Weide B. Garbe C. Neri D. Fugmann T. High-sensitivity HLA class I peptidome analysis enables a precise definition of peptide motifs and the identification of peptides from cell lines and patients' sera.Proteomics. 2017; 16: 1570-1580Crossref Scopus (44) Google Scholar, 26.Bassani-Sternberg M. Bräunlein E. Klar R. Engleitner T. Sinitcyn P. Audehm S. Straub M. Weber J. Slotta-Huspenina J. Specht K. Martignoni M.E. Werner A. Hein R. Busch D.H. Peschel C. Rad R. Cox J. Mann M. Krackhardt A.M. Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry.Nat. Commun. 2016; 7: 13404Crossref PubMed Scopus (398) Google Scholar, 27.Sofron A. Ritz D. Neri, T. Fugmann, High-resolution analysis of the murine MHC class II immunopeptidome.Eur. J. Immunol. 2016; 46: 319-328Crossref PubMed Scopus (33) Google Scholar). A similar tool, MixMHCp (22.Bassani-Sternberg M. Gfeller D. Unsupervised HLA peptidome deconvolution improves ligand prediction accuracy and predicts cooperative effects in peptide-HLA interactions.J. Immunol. 2016; 197: 2492-2499Crossref PubMed Scopus (72) Google Scholar) has been applied to the deconvolution of MHC class I EL data with performance comparable to GibbsCluster. However, neither of these methods can fully deconvolute the complete number of MHC specificities present in each data set, especially for cell lines containing overlapping binding motifs and/or lowly expressed molecules (as in the case of HLA-C). Moreover, for both methods the association of each of the clustered solutions to a specific HLA molecule must be guided by prior knowledge of the MHC binding motifs, for instance by recurring to MHC-peptide binding predictions (16.Nielsen M. Connelley T. Ternette N. Improved prediction of bovine leucocyte antigens (BoLA) presented ligands by use of mass-spectrometry-determined ligand and in vitro binding data.J. Proteome Res. 2018; 17: 559-567Crossref PubMed Scopus (26) Google Scholar). Therefore, both methods require some degree of manual intervention for deconvolution and allele annotation. A recently published method was suggested to overcome this limitation. The computational framework by Bassani-Sternberg et al. (28.Bassani-Sternberg M. Chong C. Guillaume P. Solleder M. Pak H. Gannon P.O. Kandalaft L.E. Coukos G. Gfeller D. Deciphering HLA-I motifs across HLA peptidomes improves neo-antigen predictions and identifies allostery regulating HLA specificity.PLOS Computational Biol. 2017; 13: e1005725Crossref PubMed Scopus (110) Google Scholar) employs MixMHCp (22.Bassani-Sternberg M. Gfeller D. Unsupervised HLA peptidome deconvolution improves ligand prediction accuracy and predicts cooperative effects in peptide-HLA interactions.J. Immunol. 2016; 197: 2492-2499Crossref PubMed Scopus (72) Google Scholar) to generate peptide clusters and binding motifs for a panel of poly-specificity MS data sets, and next links each cluster to an HLA molecule based on allele co-occurrence and exclusion principles. Although this approach constitutes a substantial step forward for aiding the interpretation of MS EL data, it has some substantial shortcomings. First and foremost, the success of the method is tied to the ability of MixMHCp to identify all the binding motifs in a given MS data set, an ability that is challenged in particular for cell lines containing MHC alleles with similar binding motifs, and for molecules characterized by low expression levels (22.Bassani-Sternberg M. Gfeller D. Unsupervised HLA peptidome deconvolution improves ligand prediction accuracy and predicts cooperative effects in peptide-HLA interactions.J. Immunol. 2016; 197: 2492-2499Crossref PubMed Scopus (72) Google Scholar, 29.Alvarez B. Barra C. Nielsen M. Andreatta M. Computational tools for the identification and interpretation of sequence motifs in immunopeptidomes.Proteomics. 2018; 18: e1700252Crossref PubMed Scopus (29) Google Scholar). Secondly, successful HLA labeling of the obtained clusters is limited by allele co-occurrences and exclusions across multiple cell line data sets. Although one may argue that this shortcoming is destined to wane as more immunopeptidomics data sets are accumulated in public databases, there currently remain multiple cases when co-occurrence and exclusion principles fail to completely deconvolute peptidome specificities (28.Bassani-Sternberg M. Chong C. Guillaume P. Solleder M. Pak H. Gannon P.O. Kandalaft L.E. Coukos G. Gfeller D. Deciphering HLA-I motifs across HLA peptidomes improves neo-antigen predictions and identifies allostery regulating HLA specificity.PLOS Computational Biol. 2017; 13: e1005725Crossref PubMed Scopus (110) Google Scholar). Inspired by the framework outlined by Bassani-Sternberg et al. (28.Bassani-Sternberg M. Chong C. Guillaume P. Solleder M. Pak H. Gannon P.O. Kandalaft L.E. Coukos G. Gfeller D. Deciphering HLA-I motifs across HLA peptidomes improves neo-antigen predictions and identifies allostery regulating HLA specificity.PLOS Computational Biol. 2017; 13: e1005725Crossref PubMed Scopus (110) Google Scholar) and by the earlier success of the pan-specific NNAlign framework for modeling peptide-MHC binding (14.Jurtz V. Paul S. Andreatta M. Marcatili P. Peters B. Nielsen M. NetMHCpan-4.0: improved peptide-MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data.J. Immunol. 2017; 199: 3360-3368Crossref PubMed Scopus (672) Google Scholar), we here present a novel machine learning algorithm resolving these shortcomings, enabling a fully automated clustering and labeling of MS EL data. The algorithm is an extension of the NNAlign neural network framework (15.Nielsen M. Andreatta M. NNAlign: a platform to construct and evaluate artificial neural network models of receptor-ligand interactions.Nucleic Acids Res. 2017; 45: W344-W349Crossref PubMed Scopus (33) Google Scholar, 30.Nielsen M. Lund O. NN-align. An artificial neural network-based alignment algorithm for MHC class II peptide binding prediction.BMC Bioinformatics. 2009; 10: 296Crossref PubMed Scopus (375) Google Scholar, 31.Andreatta M. Schafer-Nielsen C. Lund O. Buus S. Nielsen M. NNAlign: A web-based prediction method allowing non-expert end-user discovery of sequence motifs in quantitative peptide data.PLOS ONE. 2011; 6: e26781Crossref PubMed Scopus (48) Google Scholar), and is capable of taking a mixed training set composed of single-allele (SA) data (peptides assigned to single MHCs) and multi-allele (MA) data (peptides with multiple options for MHCs assignments) as input and deconvolute the individual MHC restriction of all MA peptides while learning the binding specificities of all the MHCs present in the training set. Compared with earlier approaches for peptidome deconvolution, annotation, and prediction model training (e.g. GibbsCluster/NNAlign (29.Alvarez B. Barra C. Nielsen M. Andreatta M. Computational tools for the identification and interpretation of sequence motifs in immunopeptidomes.Proteomics. 2018; 18: e1700252Crossref PubMed Scopus (29) Google Scholar) and MixMHCp/MixMHCpred (28.Bassani-Sternberg M. Chong C. Guillaume P. Solleder M. Pak H. Gannon P.O. Kandalaft L.E. Coukos G. Gfeller D. Deciphering HLA-I motifs across HLA peptidomes improves neo-antigen predictions and identifies allostery regulating HLA specificity.PLOS Computational Biol. 2017; 13: e1005725Crossref PubMed Scopus (110) Google Scholar)), NNAlign_MA performs these three tasks simultaneously, by iteratively updating the clustering, MHC annotation and peptide binding predictions in an integrated framework. NNAlign_MA does not require manual curation to assign the correct number of clusters, nor for the annotation of clusters to their respective MHC molecule. NNAlign_MA is available at: www.cbs.dtu.dk/suppl/immunology/NNAlign_MA/NNAlign_MA_testsuite.tar.gz. Several types of MHC peptide data for human (HLA) and bovine (BoLA) class I, and HLA class II were gathered to train the predictive models presented in this work. Peptide data was classified as single allele data (SA, where each peptide is associated to a single MHC restriction) and multi allele data (MA, where each peptide has multiple options for MHC restriction). MA data are generated from MS MHC ligand elution assays where most often a pan-specific antibody is applied for class I and either a pan-specific class II or a pan-DR specific antibody is applied for class II in the immuno-precipitation step leading to data sets with poly-specificities matching the MHC molecules expressed in the cell line under study. SA data were obtained from binding affinity assays, or from mass spectrometry experiments performed using genetically engineered cell lines that artificially express one single allele. HLA class I: SA data - both binding affinity (BA), and MS MHC eluted ligands (EL) - was extracted from Jurtz et al. (14.Jurtz V. Paul S. Andreatta M. Marcatili P. Peters B. Nielsen M. NetMHCpan-4.0: improved peptide-MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data.J. Immunol. 2017; 199: 3360-3368Crossref PubMed Scopus (672) Google Scholar). The MA data was collected from eight different sources (12.Bassani-Sternberg M. Pletscher-Frankild S. Jensen L.J. Mann M. Mass spectrometry of human leukocyte antigen class I peptidomes reveals strong effects of protein abundance and turnover on antigen presentation.Mol. Cell. Proteomics. 2015; 14: 658-673Abstract Full Text Full Text PDF PubMed Scopus (280) Google Scholar, 25.Ritz D. Gloger A. Weide B. Garbe C. Neri D. Fugmann T. High-sensitivity HLA class I peptidome analysis enables a precise definition of peptide motifs and the identification of peptides from cell lines and patients' sera.Proteomics. 2017; 16: 1570-1580Crossref Scopus (44) Google Scholar, 26.Bassani-Sternberg M. Bräunlein E. Klar R. Engleitner T. Sinitcyn P. Audehm S. Straub M. Weber J. Slotta-Huspenina J. Specht K. Martignoni M.E. Werner A. Hein R. Busch D.H. Peschel C. Rad R. Cox J. Mann M. Krackhardt A.M. Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry.Nat. Commun. 2016; 7: 13404Crossref PubMed Scopus (398) Google Scholar, 28.Bassani-Sternberg M. Chong C. Guillaume P. Solleder M. Pak H. Gannon P.O. Kandalaft L.E. Coukos G. Gfeller D. Deciphering HLA-I motifs across HLA peptidomes improves neo-antigen predictions and identifies allostery regulating HLA specificity.PLOS Computational Biol. 2017; 13: e1005725Crossref PubMed Scopus (110) Google Scholar, 32.Pearson H. Daouda T. Granados D.P. Durette C. Bonneil E. Courcelles M. Rodenbrock A. Laverdure J.-P. Côté C. Mader S. Lemieux S. Thibault P. Perreault C. MHC class I-associated peptides derive from selective regions of the human genome.J. Clin. Invest. 2011; 126: 4690-4701Crossref Scopus (116) Google Scholar, 33.Shraibman B. Kadosh D.M. Barnea E. Admon A. Human leukocyte antigen (HLA) peptides derived from tumor antigens induced by inhibition of DNA methylation for development of drug-facilitated immunotherapy.Mol. Cell. Proteomics. 2016; 15: 3058-3070Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar, 34.Mommen G.P.M. Frese C.K. Meiring H.D. van Gaans-van den Brink J. de Jong A.P.J.M. van Els C.A.C.M. Heck A.J.R. Expanding the detectable HLA peptide repertoire using electron-transfer/higher-energy collision dissociation (EThcD).Proc. Natl. Acad. Sci. U.S.A. 2014; 111: 4507-4512Crossref PubMed Scopus (140) Google Scholar, 35.Gloger A. Ritz D. Fugmann T. Neri D. Mass spectrometric analysis of the HLA class I peptidome of melanoma cell lines as a promising tool for the identification of putative tumor-associated HLA epitopes.Cancer Immunol. Immunother. 2016; 65: 1377-1393Crossref PubMed Scopus (30) Google Scholar). Both data sets were filtered to include only peptides of length 8–14 amino acids. Additional information concerning the HLA class I MA data can be found in supplemental Table S1 and information concerning the SA BA and EL data sets in supplemental Table S2. HLA-II: BA data was extracted from the NetMHCIIpan-3.2 publication (36.Jensen K.K. Andreatta M. Marcatili P. Buus S. Greenbaum J.A. Yan Z. Sette A. Peters B. Nielsen M. Improved methods for predicting peptide binding affinity to MHC class II molecules.Immunology. 2018; 154: 394-406Crossref PubMed Scopus (368) Google Scholar). As for EL data, the Immune Epitope Database (37.Vita R. Mahajan S. Overton J.A. Dhanda S.K. Martini S. Cantrell J.R. Wheeler D.K. Sette A. Peters B. The Immune Epitope Database (IEDB): 2018 update.Nucleic Acids Res. 2019; 47: D339-D343Crossref PubMed Scopus (710) Google Scholar) (IEDB) was queried to identify publications with a large number of allele annotated EL data, both SA and MA (27.Sofron A. Ritz D. Neri, T. Fugmann, High-resolution analysis of the murine MHC class II immunopeptidome.Eur. J. Immunol. 2016; 46: 319-328Crossref PubMed Scopus (33) Google Scholar, 38.Clement C.C. Becerra A. Yin L. Zolla V. Huang L. Merlin S. Follenzi A. Shaffer S.A. Stern L.J. Santambrogio L. The dendritic cell major histocompatibility complex II (MHC II) peptidome derives from a variety of processing pathways and includes peptides with a broad spectrum of HLA-DM sensitivity.J. Biol. Chem. 2016; 291: 5576-5595Abstract Full Text Full Text PDF PubMed Scopus (39) Google Scholar, 39.Bergseng E. Dørum S. Arntzen, M. Nielsen, S. Nygård, S. Buus, G. A. de Souza, L. M. Sollid. Different binding motifs of the celiac disease-associated HLA molecules DQ2.5, DQ2.2, and DQ7.5 revealed by relative quantitative proteomics of endogenous peptide repertoires.Immunogenetics. 2015; 67: 73-84Crossref PubMed Scopus (64) Google Scholar, 40.Heyder T. Kohler M. Tarasova N.K. Haag S. Rutishauser D. Rivera N.V. Sandin C. Mia S. Malmström V. Wahlström J. Holmdahl R. Eklund A. Zubarev R.A. Grunewald J. Ytterberg A.J. Approach for identifying human leukocyte antigen (HLA)-DR bound peptides from scarce clinical samples.Mol. Cell. Proteomics. 2016; 15: 3017-3029Abstract Full Text Full Text PDF PubMed Scopus (34) Google Scholar, 41.Nelde A. Kowalewski D.J. Backert L. Schuster H. Werner J.-O. Klein R. Kohlbacher O. Kanz L. Salih H.R. Rammensee H.-G. Stevanović S. Walz J.S. HLA ligandome analysis of primary chron