T-cell receptor binding prediction - A machine learning revolution

Aug 5, 2026

Recent advances in immune sequencing and experimental techniques are producing large T cell receptor (TCR) repertoire datasets, enabling computational models to predict TCR binding specificity. Despite the immense diversity of TCRs and epitopes, progress continues across unsupervised clustering, supervised learning, and modern Protein Language Models (PLMs). This review surveys key model families, analyzes recurring challenges such as limited generalization to unseen epitopes, dataset biases, and validation shortcomings, and highlights interpretability needs for large black-box models.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages34

Reading Time1 h 26 min

AccessFree to download

Background

Adaptive immunity and TCR recognition

TCR diversity and pMHC interaction

T cell receptor specificity prediction

Importance for immunotherapy development

Generalization challenge to novel epitopes

Limitations of available datasets

Key databases and emerging datasets

Paired-chain data constraints

arXiv :2312 . 16594v2 [ q-bio .QM] 23 Jul 2024 T-cell receptor binding prediction: A machine learning revolution Anna Weber* Aurélien Pélissier† María Rodríguez Martínez‡ § July 24, 2024 Abstract Recent advancements in immune sequencing and experimental techniques are generating extensive T cell receptor (TCR) repertoire data, enabling the development of models to predict TCR binding specificity. Despite the computational challenges posed by the vast diversity of TCRs and epitopes, significant progress has been made. This review explores the evolution of computational models designed for this task, emphasizing machine learning efforts, including early unsupervised clustering approaches, supervised models, and recent applications of Protein Language Models (PLMs), deep learning models pretrained on extensive collections of unlabeled protein sequences that capture crucial biological properties. We survey the most prominent models in each category and offer a critical discussion on recurrent challenges, including the lack of generalization to new epitopes, dataset biases, and shortcomings in model validation designs. Focusing on PLMs, we discuss the transformative impact of Transformer-based protein models in bioinformatics, particularly in TCR specificity analysis. We discuss recent studies that exploit PLMs to deliver notably competitive performances in TCR-related tasks, while also examining current limitations and future directions. Lastly, we address the pressing need for improved interpretability in these often opaque models, and examine current efforts to extract biological insights from large black box models. Keyword: Machine Learning; T cell Receptor; Specificity Prediction; Protein Language Models; Interpretability. *IBM Research Europe, 8803 Rüschlikon, Switzerland. †Institute of Computational Life Sciences, Zürich University of Applied Sciences (ZHAW), 8820 Wädenswil, Switzerland. ‡Biomedical Informatics and Data Science, Yale School of Medicine, New Haven, CT 06510, USA. § [Correspondence to maria.rodriguezmartinez@yale.edu](Correspondence to maria.rodriguezmartinez@yale.edu). 1 Background T cells are an essential component of the adaptive immune system due to their ability to orchestrate targeted, effective immune responses through cell-based and cytokinerelease mechanisms. While T cell functions are diverse, their activation, differentiation, proliferation, and function are all governed by their T cell receptors (TCR), which enable them to recognize non-self antigens arising from infectious agents or diseased cells [1] . To face a diverse and ever-evolving array of antigens, the immune system has evolved the capability to generate a huge array of distinct TCRs. This diversity is achieved through a random process of DNA rearrangement, which involves the recombination of the germline V, D, and J gene segments and the deletion and insertion of nucleotides at the V(D)J junctions. While the theoretical diversity of different TCRs is estimated to be as high as 1019 [2], the realized diversity in an individual is much smaller, typically ranging between 106 and 1010 [3] . At the molecular level, TCRs interact with peptides presented on the major histocompatibility complex (MHC), a complex commonly referred to as pMHC. Although the interaction between pMHC and TCR is highly specific, a single TCR can often recognize multiple pMHC complexes. Indeed, some TCRs have been shown to recognize up to a million different epitopes [4] . This multivalency is necessary to ensure that the realized diversity in one individual can recognize a significantly broader array of potential antigens. 2 T cell receptor specificity prediction. The accurate prediction of TCR-pMHC binding is crucial for accurately estimating immune responses and holds the promise to revolutionize the development of immunotherapies. For instance, the precise determination of the epitopes recognized by expanded TCR clones can aid in

FAQs

Why is TCR binding specificity prediction important for immunotherapy research?

Accurate prediction helps estimate immune responses and supports development of more effective, less toxic immunotherapies. It also aids epitope identification, auto-antigen discovery in autoimmune disease, and investigation of pathogens driving T-cell responses.

What makes generalizing TCR specificity models to novel epitopes difficult?

Models often struggle to generalize because experimentally validated TCR-epitope interaction datasets are scarce and the diversity of epitopes sampled is limited. This leads to difficulty covering the broader epitope space encountered in practice.

What role do Protein Language Models (PLMs) play in recent TCR prediction efforts?

PLMs, especially Transformer-based protein models trained on large unlabeled sequence collections, capture biological properties useful for TCR specificity analysis. The review discusses studies where PLM-based approaches achieve competitive performance, while noting current limitations.

Aug 5, 2026

Upload by Genevieve