Functional evaluation of TCR-pMHC pairs at scale allows in silico TCR reactivity prediction.

AlphaFold3 structural predictions accurately identify reactive TCR-pMHC pairs without task-specific training, challenging previous reactivity estimates in immunology.

• Why it matters: Accurate prediction of TCR reactivity is essential for advancing immunological research and developing personalized immunotherapies, yet current models may overestimate reactivity.
• What they did: Researchers developed T-RAP to generate large TCR libraries and evaluated thousands of TCR-pMHC pairs under standardized conditions, revealing only about 50% of previously reported reactive TCRs.
• The result: AlphaFold3 successfully predicted reactive TCR-pMHC pairs and can be used to shortlist TCRs targeting patient-specific cancer neoantigens, enhancing in silico reactivity prediction feasibility.