New Benchmarking Shows Limited Generalization Power of TCR Antigenic Epitope Prediction Models

2026-06-03

AI Summary

Current T cell receptor (TCR) epitope prediction models suffer from limited generalizability and a lack of rigorously defined unseen benchmark datasets for unbiased evaluation. This work addresses these gaps by constructing, for the first time, two complementary and strictly partitioned unseen benchmark datasets, thereby establishing a standardized evaluation framework for TCR–antigen binding prediction. By integrating immune repertoire analysis, computational immunology, and machine learning evaluation methodologies, the study systematically uncovers the performance bottlenecks of state-of-the-art models in real-world scenarios. The resulting benchmarks and foundational framework provide a reliable basis for the development and assessment of next-generation prediction algorithms.

Abstract

Accurate computational prediction of T cell receptor (TCR) antigen specificity would transform the study of T cell biology and enable scalable immune engineering, yet existing models lack sufficient sensitivity and specificity for broad applications. A major limitation is the absence of rigorously defined, unseen benchmark datasets that allow unbiased evaluation of model performance and generalizability. Here, we describe two complementary classes of datasets that meet this criterion and argue that they provide both a robust framework for model assessment and a foundation for next-generation TCR-antigen prediction algorithm development.

Problem

• TCR
• antigen specificity
• benchmarking
• generalization
• epitope prediction

Innovation

• TCR-antigen prediction
• benchmark dataset
• model generalization
• immune engineering
• computational immunology