TCR-BERT: learning the grammar of T-cell receptors for flexible antigen-xbinding analyses
TCR-BERT: изучение «грамматики» T-клеточных рецепторов для гибкого анализа связывания антигенов
2021-11-20
SCID: 54.1/tncqp8e5
Discuss with AI
T-cell receptor sequencesTCR sequence clusteringTCR-BERTTCR-antigen binding predictionself-supervised transfer learning
Figures from the paper
Abstract (AI)
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. We demonstrate that TCR-BERT can be used to build state-of-the-art TCR-antigen binding predictors with improved generalizability compared to prior methods. TCR-BERT simultaneously facilitates clustering sequences likely to share antigen specificities. It also facilitates 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. TCR-BERT can be a useful tool for T-cell scientists, enabling greater understanding and more diverse applications, and provides a conceptual framework for leveraging unlabeled data to improve machine learning on biological sequences.
Key Findings
1
TCR-BERT focuses on biologically significant residues, providing a general framework for using unlabeled biological sequence data.
2
TCR-BERT supports clustering TCR sequences that are likely to share antigen specificities.
3
TCR-BERT uses self-supervised transfer learning on unlabeled TCR sequences to learn versatile sequence representations.
4
The learned representations enable state-of-the-art TCR–antigen binding predictors with improved generalizability over prior methods.
5
The model facilitates computational design of novel TCR sequences with engineered binding affinities, addressing an unsolved problem.
Research Object
T-cell receptor (TCR) sequences and their interactions with antigens
Research Subject
antigen-binding specificity, sequence representation, clustering, and engineered binding-affinity design of TCRs
Publication Details
Publication Date
2021-11-20
Journal
Publisher
ISSN
Cited by
85
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai11
Scikit-learn: Machine Learning in Python2012
SciPy 1.0: fundamental algorithms for scientific computing in Python2020
The Protein Data Bank2000
PyTorch: An Imperative Style, High-Performance Deep Learning Library2019
Array programming with NumPy2020
Neural Machine Translation by Jointly Learning to Align and Translate2014
MUSCLE: a multiple sequence alignment method with reduced time and space complexity2004
Transformers: State-of-the-Art Natural Language Processing2020
DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome2021
Transformer neural network for protein-specific de novo drug generation as a machine translation problem2021
TCRMatch: Predicting T-Cell Receptor Specificity Based on Sequence Similarity to Previously Characterized Receptors2021