Novel modularity-based t-cell receptor specificity detection algorithm 3830

Description Mapping T-Cell Receptors (TCRs) to specific antigens is a key challenge in understanding the functional implications of TCR diversity and specificity when combining TCR sequencing data with other biological and clinical aspects for actionable insights in precision medicine. Here we present a novel modularity optimization based method that utilizes the concept of split-penalty in graph theoretical algorithms to identify local TCR antigen-specificity groups in active immune cell repertoires. For modeling single-sequence input TCR CDR3 regions, we developed a transformer based feature space representation and utilize this latent space for downstream specificity group identification. Our method integrates the TCR analysis with single cell RNA sequencing derived phenotypes of top expanded TCR clones as well as repertoire-level comprehensive single cell immune profiling. Funding Sources Supported by OU-BIC2.0; NSF- 2341725; NIH-NCI K25CA270079 Topic Categories Computational and Systems Immunology (COMP)