HLA Typing Prediction and Cross-Reactivity Network Mapping via HLA-Associated TCR Cluster Integration 2321

Abstract Description The composition of human leukocyte antigens (HLA) in individuals shapes their T cell receptor (TCR) repertoire through thymic selection and restricted peptide-HLA (pHLA) pools. While cases of TCR-HLA cross-relativity have been reported decades ago, the extent of this cross-reactivity among different HLA alleles remains poorly understood. Here, we introduce THmap, a TCR-based HLA similarity mapping network, designed to quantify TCR-HLA cross-reactivity. We conducted TCR cluster-based HLA enrichment analysis and identified over three million HLA-associated TCRs. This extensive dataset allowed us to construct accurate HLA inference models based on one’s TCR repertoires. Additionally, we created similarity networks for both class I and class II HLA alleles, illustrating how peptide cross-presentation contributes to TCR-HLA cross-reactivity. Using THmap, we clustered populations from various ethnic groups based on their HLA compositions. This analysis revealed novel HLA disease susceptibilities that single-HLA enrichment analyses overlooked, especially in underrepresented groups like Black populations. Furthermore, THmap can prioritize optimal HLA mismatch candidates for organ transplantation, thereby improving patient survival rates. In sum, THmap is a comprehensive model that bridges TCR-HLA associations, offering valuable insights into TCR-HLA cross-reactivity and its implications for disease susceptibility and organ transplantation strategies. Funding Sources NCI R01CA258524 Topic Categories Computational and Systems Immunology (COMP)