EUVIMEDEUVIMED entdecken
Open Access

Predicting specificity of TCR-pMHC interactions using machine-learning and biophysical models

Martin Culka, Nicolas W Lounsbury, William Thrift, Santrupti Nerli, Andrew Wallace, Gergő Nikolényi, Darya Orlova, Kiran Mukhyala, Mohammed AlQuraishi

Cell Systems · 2026-09 · 17 (9) · 101700

DOI: 10.1016/j.cels.2026.101700 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Cell Systems, 2026-09. Verfasst von Martin Culka und 8 weiteren Autor:innen. Beteiligte Einrichtung: Department of Systems Biology, Columbia University, New York, NY 10032, USA. Hauptthemen laut MEDLINE: Receptors, Antigen, T-Cell. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 66 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NLM NIH HHS, NIGMS NIH HHS, National Library of Medicine.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Understanding T cell receptor (TCR) discrimination of MHC-presented epitope peptides (pMHCs) remains challenging. While machine-learning (ML)-based predictions of TCR specificity have gained attention, their capacity to generalize to unseen peptides is often misinterpreted. Using a proprietary cancer patient dataset, we show that ML methods succeed in predicting TCR specificity for known peptides but fail to generalize to novel peptides. Conversely, physics-based methods outperform ML methods on novel peptides but underperform on known peptides. In light of these observations, we develop a new ML method that leverages protein foundation models to achieve better or comparable performance than existing ML and biophysical methods on both in- and out-of-distribution TCR-pMHC specificity prediction. We furthermore characterize method performance as a function of distance of TCR sequence specificity between training and test sets. Our analysis elucidates the current limitations of modeling TCR-pMHC interactions and outlines new avenues for method development and data acquisition.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Receptors, Antigen, T-CellMachine LearningHumansPredictive Learning ModelsPeptidesEpitopes, T-LymphocytePrediction Algorithms

Stichwörter der Autor:innen

AlphaFoldESMT-cell specificityTCR pMHCcancer neoantigensexplainable MLforce fieldmachine learningprotein-peptideprotein-protein

Substanzen

Receptors, Antigen, T-CellPeptidesEpitopes, T-Lymphocyte

Autor:innen und Institutionen

  • Department of Systems Biology, Columbia University, New York, NY 10032, USA.
  • Genentech, South San Francisco, CA 94080, USA.
  • Genentech, South San Francisco, CA 94080, USA. Electronic address: mukhyala@gene.com.
  • Department of Systems Biology, Columbia University, New York, NY 10032, USA. Electronic address: ma4129@cumc.columbia.edu.

Förderung

  • NLM NIH HHS
  • NIGMS NIH HHS
  • National Library of Medicine
  • Genentech
  • National Institute of General Medical Sciences

Zitieren

APA

Culka, M., Lounsbury, N. W., Thrift, W., Nerli, S., Wallace, A., Nikolényi, G., Orlova, D., Mukhyala, K., & AlQuraishi, M. (2026). Predicting specificity of TCR-pMHC interactions using machine-learning and biophysical models. Cell Systems, 17(9), 101700. https://doi.org/10.1016/j.cels.2026.101700

Vancouver

Culka M, Lounsbury NW, Thrift W, Nerli S, Wallace A, Nikolényi G, et al. Predicting specificity of TCR-pMHC interactions using machine-learning and biophysical models. Cell Systems. 2026;17(9):101700. doi:10.1016/j.cels.2026.101700

BibTeX
@article{Culka2026,
  author = {Culka, Martin and Lounsbury, Nicolas W and Thrift, William and Nerli, Santrupti and Wallace, Andrew and Nikolényi, Gergő and Orlova, Darya and Mukhyala, Kiran and AlQuraishi, Mohammed},
  title = {Predicting specificity of TCR-pMHC interactions using machine-learning and biophysical models},
  journal = {Cell Systems},
  year = {2026},
  volume = {17},
  number = {9},
  pages = {101700},
  doi = {10.1016/j.cels.2026.101700}
}
Kostenlos starten →