Rapid Assessment of T-Cell Receptor Specificity of the Immune Repertoire

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• Accurate assessment of T-cell receptor (TCR)-antigen specificity is crucial for advancing cancer immunotherapy.
• Existing predictive models lack the high-throughput capacity for assessing TCR-peptide pairs.
• Advances in sequencing and crystallography provide rich data for studying TCR-peptide-MHC interactions.

Purpose of the Study:

• To introduce RACER, a novel pairwise energy model for rapid assessment of TCR-peptide affinity at the immune repertoire level.
• To develop a supervised machine learning approach for efficiently distinguishing strong TCR-peptide binding pairs from weak ones.
• To enable physical interpretation of interaction patterns within TCR-p-MHC systems.

Main Methods:

• Development of RACER, a supervised machine learning model based on a pairwise energy approach.
• Training the model on existing data to learn TCR-peptide binding characteristics.
• Application of RACER to simulate thymic selection and estimate recognition rates for various peptides.

Main Results:

• RACER efficiently and accurately resolves strong TCR-peptide binding pairs from weak ones.
• The model's trained parameters offer physical insights into TCR-p-MHC system interactions.
• RACER accurately estimates recognition rates for tumor-associated neoantigens and foreign peptides in simulated thymic selection.

Conclusions:

• RACER provides a powerful computational tool for high-throughput assessment of TCR-peptide affinity.
• The model aids in identifying tumor antigen-specific T-cells within an individual patient's immune repertoire.
• RACER has significant utility for advancing personalized cancer immunotherapy research.