Unsupervised identification of low-frequency antigen-specific TCRs using distance-based anomaly scoring.

Abstract

Identifying antigen-specific T cell receptors (TCRs) within the diverse human repertoire remains challenging, particularly for low-frequency clonotypes. Here, we present TCR-RADAR (TCR Rare Antigen-specific Detection by Anomaly Ranking), an unsupervised approach that detects low-frequency antigen-specific TCRs through distance-based anomaly detection in TCR sequence space. Using TCRdist3 to quantify sequence distances, we identify query TCRs that are anomalous relative to reference repertoires within their V-J gene combinations. We validated this approach across three immunological contexts: COVID-19 infection, influenza vaccination, and yellow fever vaccination. For SARS-CoV-2-specific TCR detection in a COVID-19 patient, our method achieved 34.3% precision, substantially higher than similarity-based (ALICE: 8.0%) and frequency-based methods (edgeR: 5.8%, the Pogorelyy method: 6.3%), and uniquely detected low-frequency antigen-specific TCRs present at a clone count of 1. The minimal overlap with conventional approaches (0%-6.7% across the three datasets) indicates our method captures distinct TCR clones overlooked by existing analyses. This distance-based approach provides a complementary strategy for TCR specificity detection, particularly valuable for identifying rare antigen-specific clones essential for understanding immune responses.