TCRmodel2: high-resolution modeling of T cell receptor recognition using deep learning.
Rui Yin1,2, Helder V Ribeiro-Filho1,3, Valerie Lin1,4
1University of Maryland Institute for Bioscience and Biotechnology Research, Rockville, MD 20850, USA.
Nucleic Acids Research
|May 4, 2023
Summary
A new web server, TCRmodel2, models T-cell receptor (TCR) and peptide-MHC complexes from sequence. This computational tool aids understanding of immunity and accelerates vaccine and immunotherapeutic design.
Area of Science:
- Immunology
- Structural Biology
- Bioinformatics
Background:
- The cellular immune system relies on T-cell receptors (TCRs) to recognize peptide-MHC complexes.
- Understanding TCR-peptide-MHC interactions is crucial for immunity, vaccine design, and immunotherapeutics.
- Experimental structure determination is limited, necessitating computational modeling.
Purpose of the Study:
- To update the TCRmodel web server for modeling TCR-peptide-MHC complexes from sequence.
- To provide an accurate and efficient computational tool for structural analysis of TCR-peptide-MHC interactions.
Main Methods:
- Adaptations of AlphaFold were utilized to develop the TCRmodel2 computational approach.
- The method models TCR-peptide-MHC complexes directly from amino acid sequences.
- Benchmarking was performed to assess the accuracy of the generated models.
Main Results:
- TCRmodel2 demonstrates accuracy comparable to or exceeding other methods for modeling TCR-peptide-MHC complexes.
- The web server provides user-friendly sequence submission and rapid model generation (within 15 minutes).
- Output includes confidence scores and an integrated molecular viewer for model assessment.
Conclusions:
- TCRmodel2 offers a significant advancement in computational modeling of TCR-peptide-MHC interactions.
- The tool facilitates structural insights into immunity and aids in the development of novel vaccines and therapeutics.
- The web server is publicly accessible at https://tcrmodel.ibbr.umd.edu.
