The accurate prediction of peptide-major histocompatibility complex (MHC) class I binding probabilities is a critical endeavor in immunoinformatics, with broad implications for vaccine development and immunotherapies. While recent deep neural network based approaches have showcased promise in peptide-MHC (pMHC) prediction, they have two shortcomings: (i) they rely on hand-crafted pseudo-sequence extraction, (ii) they do not generalize well to different datasets, which limits the practicality of these approaches. While existing methods rely on a 34 amino acid pseudo-sequence, our findings uncover the involvement of 147 positions in direct interactions between MHC and peptide. We further show that neural architectures can learn the intricacies of pMHC binding using even full sequences. To this end, we present PerceiverpMHC that is able to learn accurate representations on full-sequences by leveraging efficient transformer based architectures. Additionally, we propose IMGT/RobustpMHC that harnesses the potential of unlabeled data in improving the robustness of pMHC binding predictions through a self-supervised learning strategy. We extensively evaluate RobustpMHC on eight different datasets and showcase an overall improvement of over 6% in binding prediction accuracy compared to state-of-the-art approaches. We compile CrystalIMGT, a crystallography-verified dataset presenting a challenge to existing approaches due to significantly different pMHC distributions. Finally, to mitigate this distribution gap, we further develop a transfer learning pipeline.

IMGT, the international ImMunoGeneTics information system (http://imgt.cines.fr), is the reference in immunogenetics and immunoinformatics. IMGT standardizes and manages the complex immunogenetic data that include the immunoglobulins (IG) or antibodies, the T cell receptors (TR), the major histocompatibility complex (MHC) and the related proteins of the immune system (RPI), which belong to the immunoglobulin superfamily (IgSF) and the MHC superfamily (MhcSF). The accuracy and consistency of IMGT data and the coherence between the different IMGT components (databases, tools and Web resources) are based on IMGT-ONTOLOGY, the first ontology for immunogenetics and immunoinformatics. IMGT-ONTOLOGY manages the immunogenetics knowledge through diverse facets relying on seven axioms, 'IDENTIFICATION', 'DESCRIPTION', 'CLASSIFICATION', 'NUMEROTATION', 'LOCALIZATION', 'ORIENTATION' and 'OBTENTION', that postulate that objects, processes and relations have to be identified, described, classified, numerotated, localized, orientated, and that the way they are obtained has to be determined. These axioms constitute the Formal IMGT-ONTOLOGY, also designated as IMGT-Kaleidoscope. These axioms have been essential for the conceptualization of the molecular immunogenetics knowledge and for the creation of IMGT. Indeed all the components of the IMGT integrated system have been developed, based on standardized concepts and relations, thus allowing IMGT to bridge biological and computational spheres in bioinformatics. The same axioms can be used to generate concepts for multi-scale level approaches at the molecule, cell, tissue, organ, organism or population level, emphasizing the generalization of the application domain. In that way the Formal IMGT-ONTOLOGY represents a paradigm for the elaboration of ontologies in system biology.

MOTIVATION: IMGT, the international ImMunoGeneTics database (http:@imgt.cines.fr:8104), created by M.-P. Lefranc, is an integrated database specializing in antigen receptors (immunoglobulins and T-cell receptors) and major histocompatibility complex (MHC) of all vertebrate species. IMGT accurate immunogenetics data are based on the standardization of the biological knowledge provided by the 'ImMunoGeneTics' IMGT-ONTOLOGY. The IMGT-ONTOLOGY describes the classification and specification of terms needed for immunogenetics and bioinformatics. IMGT-ONTOLOGY covers four main concepts: 'IDENTIFICATION', 'DESCRIPTION', 'CLASSIFICATION' and 'OBTENTION'. These concepts allow an extensive and standardized description and characterization of immunoglobulin and T-cell receptor data. The controlled vocabulary and the annotation rules are indispensable to ensure accuracy, consistency and coherence in IMGT. IMGT-ONTOLOGY allows scientists and clinicians to use, for the first time, identical terms with the same meaning in immunogenetics. It provides a semantic repository that will improve interoperability between specialist and generalist databases.