Related Experiment Video
Updated: Jun 26, 2026
08:27
Detection of Human Leukocyte Antigen Biomarkers in Breast Cancer Utilizing Label-free Biosensor Technology
Published on: March 24, 2015
15.3K
THLANet: A deep learning framework for predicting TCR-pHLA binding in immunotherapy applications
Xu Long1, Qiang Yang1, Weihe Dong1
1School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China.
Plos Computational Biology
|September 12, 2025
Summary
Predicting T-cell receptor (TCR) binding to neoantigens is key for cancer immunotherapy. THLANet, a deep learning model, accurately predicts TCR-neoantigen interactions using sequence data, advancing anti-tumor immunity research.
Area of Science:
- Immunology
- Computational Biology
- Bioinformatics
Background:
- Adaptive immunity is crucial for anti-tumor responses, relying on T-cell receptor (TCR) recognition of tumor antigens presented by human leukocyte antigens (HLA).
- The limited ability of TCRs to recognize all potential neoantigens poses a challenge for effective cancer immunotherapy.
- Accurate prediction of TCR-neoantigen binding is essential for assessing immunogenicity and guiding therapeutic strategies.
Purpose of the Study:
- To develop a deep learning model, THLANet, for predicting the binding specificity between TCRs and neoantigens presented by class I HLA molecules.
- To enhance sequence feature representation using evolutionary scale modeling-2 (ESM-2) for improved prediction accuracy.
- To provide insights into the structural basis of TCR-antigen interactions.
Main Methods:
- Developed THLANet, a deep learning model utilizing ESM-2 for sequence feature extraction.
- Constructed a TCR-pHLA binding database using scTCR-seq data to train and validate the model.
- Evaluated model performance on clinical cancer data across diverse cancer types.
- Analyzed complementarity-determining region (CDR3) sequences and performed alanine scanning simulations.
Main Results:
- THLANet accurately predicts TCR-neoantigen binding specificity using only TCR CDR3β, antigen, and class I HLA sequence information.
- The model demonstrates clinical potential validated on scTCR-seq data and diverse cancer types.
- Analysis provided novel insights into the 3D binding interactions between TCRs and antigens.
Conclusions:
- THLANet offers a robust and accurate method for predicting TCR-neoantigen pairing, a significant challenge in immunology.
- The model's ability to leverage sequence data simplifies prediction and offers new perspectives on TCR-antigen interactions.
- This work has implications for advancing neoantigen discovery and developing personalized cancer immunotherapies.
