TCRfinder: Improved TCR virtual screening for novel antigenic peptides with tailored language models

Yang Li,Chaoting Zhang,Xi Zhang,Yang Zhang

DOI: https://doi.org/10.1101/2024.06.27.601008

2024-07-01

Abstract:Accurate modeling of T-cell receptor (TCR) and peptide interactions is essential for immunoreaction elucidation and T-cell-based immunotherapeutic developments. We developed TCRfinder, a novel deep-learning architecture for TCR-peptide binding prediction and virtual screening. Large-scale benchmark experiments demonstrated a robust capability of TCRfinder in distinguishing interacting and non-interacting TCRs for unseen peptides, with accuracy significantly beyond current state-of-the-art methods. Furthermore, TCRfinder recognizes tumor neoantigen mutations from wild-type antigens of given TCRs, with a success rate nearly 50% higher than the best of existing methods. Detailed data analyses showed that the major advantage of TCRfinder lies in the specially trained TCR and peptide language models tailored with iterative attention network architecture, which can precisely reveal physical interaction patterns of cross-chain atoms and substantially enhance the precision of TCR-peptide interaction predictions. The open-source TCRfinder program can help facilitate large-scale deployment of high-quality TCR and neoantigen virtual screening, offering exciting potential for personalized TCR-based immunotherapies.

Bioinformatics

What problem does this paper attempt to address?

The paper primarily addresses two key issues:

1. **Improving the prediction of T-cell receptor (TCR) and peptide interactions**: A new method named TCRfinder was developed, which is a deep learning architecture used to predict the binding between T-cell receptors (TCR) and peptides, and to perform virtual screening. This method has shown a significant improvement in accuracy in distinguishing known interactions from non-interactions of TCRs with unknown peptides, demonstrating a clear advantage over existing technologies.

2. **Identifying tumor neoantigens**: A specially trained model was used to distinguish between tumor neoantigen mutations and wild-type antigens. This method can effectively identify TCRs targeting neoantigens from a given set of TCRs, with a success rate nearly 50% higher than the current best methods.

To achieve these goals, the researchers designed special TCR and peptide language models and combined them with an iterative attention network architecture to accurately reveal the physical interaction patterns between atoms across chains, thereby significantly improving the accuracy of TCR-peptide interaction predictions. Additionally, the study demonstrated the powerful capability of TCRfinder in identifying neoantigens, which is of great significance for the development of personalized TCR immunotherapies.