TCR-BERT: learning the grammar of T-cell receptors for flexible antigen-binding analyses

Kevin E. Wu, Kathryn Yost, Bence Daniel, Julia Belk, Yu Xia, Takeshi Egawa, Ansuman Satpathy, Howard Chang, James Zou
Proceedings of the 18th Machine Learning in Computational Biology meeting, PMLR 240:194-229, 2024.

Abstract

The T-cell receptor (TCR) allows T-cells to recognize and respond to antigens presented by infected and diseased cells. However, due to TCRs’ staggering diversity and the complex binding dynamics underlying TCR antigen recognition, it is challenging to predict which antigens a given TCR may bind to. Here, we present TCR-BERT, a deep learning model that applies self-supervised transfer learning to this problem. TCR-BERT leverages unlabeled TCR sequences to learn a general, versatile representation of TCR sequences, enabling numerous downstream applications. TCR-BERT can be used to build state-of-the-art TCR-antigen binding predictors with improved generalizability compared to prior methods. Simultaneously, TCR-BERT’s embeddings yield clusters of TCRs likely to share antigen specificities. It also enables computational approaches to challenging, unsolved problems such as designing novel TCR sequences with engineered binding affinities. Importantly, TCR-BERT enables all these advances by focusing on residues with known biological significance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v240-wu24b, title = {TCR-BERT: learning the grammar of T-cell receptors for flexible antigen-binding analyses}, author = {Wu, Kevin E. and Yost, Kathryn and Daniel, Bence and Belk, Julia and Xia, Yu and Egawa, Takeshi and Satpathy, Ansuman and Chang, Howard and Zou, James}, booktitle = {Proceedings of the 18th Machine Learning in Computational Biology meeting}, pages = {194--229}, year = {2024}, editor = {Knowles, David A. and Mostafavi, Sara}, volume = {240}, series = {Proceedings of Machine Learning Research}, month = {30 Nov--01 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v240/wu24b/wu24b.pdf}, url = {https://proceedings.mlr.press/v240/wu24b.html}, abstract = {The T-cell receptor (TCR) allows T-cells to recognize and respond to antigens presented by infected and diseased cells. However, due to TCRs’ staggering diversity and the complex binding dynamics underlying TCR antigen recognition, it is challenging to predict which antigens a given TCR may bind to. Here, we present TCR-BERT, a deep learning model that applies self-supervised transfer learning to this problem. TCR-BERT leverages unlabeled TCR sequences to learn a general, versatile representation of TCR sequences, enabling numerous downstream applications. TCR-BERT can be used to build state-of-the-art TCR-antigen binding predictors with improved generalizability compared to prior methods. Simultaneously, TCR-BERT’s embeddings yield clusters of TCRs likely to share antigen specificities. It also enables computational approaches to challenging, unsolved problems such as designing novel TCR sequences with engineered binding affinities. Importantly, TCR-BERT enables all these advances by focusing on residues with known biological significance. } }
Endnote
%0 Conference Paper %T TCR-BERT: learning the grammar of T-cell receptors for flexible antigen-binding analyses %A Kevin E. Wu %A Kathryn Yost %A Bence Daniel %A Julia Belk %A Yu Xia %A Takeshi Egawa %A Ansuman Satpathy %A Howard Chang %A James Zou %B Proceedings of the 18th Machine Learning in Computational Biology meeting %C Proceedings of Machine Learning Research %D 2024 %E David A. Knowles %E Sara Mostafavi %F pmlr-v240-wu24b %I PMLR %P 194--229 %U https://proceedings.mlr.press/v240/wu24b.html %V 240 %X The T-cell receptor (TCR) allows T-cells to recognize and respond to antigens presented by infected and diseased cells. However, due to TCRs’ staggering diversity and the complex binding dynamics underlying TCR antigen recognition, it is challenging to predict which antigens a given TCR may bind to. Here, we present TCR-BERT, a deep learning model that applies self-supervised transfer learning to this problem. TCR-BERT leverages unlabeled TCR sequences to learn a general, versatile representation of TCR sequences, enabling numerous downstream applications. TCR-BERT can be used to build state-of-the-art TCR-antigen binding predictors with improved generalizability compared to prior methods. Simultaneously, TCR-BERT’s embeddings yield clusters of TCRs likely to share antigen specificities. It also enables computational approaches to challenging, unsolved problems such as designing novel TCR sequences with engineered binding affinities. Importantly, TCR-BERT enables all these advances by focusing on residues with known biological significance.
APA
Wu, K.E., Yost, K., Daniel, B., Belk, J., Xia, Y., Egawa, T., Satpathy, A., Chang, H. & Zou, J.. (2024). TCR-BERT: learning the grammar of T-cell receptors for flexible antigen-binding analyses. Proceedings of the 18th Machine Learning in Computational Biology meeting, in Proceedings of Machine Learning Research 240:194-229 Available from https://proceedings.mlr.press/v240/wu24b.html.

Related Material