LANTERN: TCR-peptide binding prediction via large language model representations
PeerJ · 30.03.2026 · 14 · e20980
DOI: 10.7717/peerj.20980 ↗Kurz gefasst
Fachartikel in der Zeitschrift PeerJ, online erschienen am 30.03.2026. Verfasst von Cong Qi und 4 weiteren Autor:innen. Beteiligte Einrichtung: Computer Science, New Jersey Institute of Technology, Newark, New Jersey, United States. Hauptthemen laut MEDLINE: Receptors, Antigen, T-Cell, Peptides. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 44 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIGMS NIH HHS.
Abstract (Originalsprache)
Predicting T-cell receptor (TCR) and peptide-major histocompatibility complex (pMHC) interactions is critical for advancing targeted immunotherapies and personalized medicine. However, existing models often struggle with limited labeled data and poor generalization to novel epitopes. We present LANTERN (Large lAnguage model-powered TCR-Enhanced Recognition Network), a novel deep learning framework that combines pretrained protein and molecular language models with a cross-modality fusion mechanism. Specifically, LANTERN encodes TCR sequences using ESM and peptides as Simplified Molecular Input Line Entry System (SMILES) strings via MolFormer, capturing both evolutionary and chemical properties. A Multi-Head Cross-Attention (MHCA) module is introduced to align TCR and peptide representations, enabling the model to focus on interaction-relevant features across domains. This architecture improves generalization in zero-shot and few-shot scenarios. Extensive experiments on the TCHard benchmark demonstrate that LANTERN achieves competitive and robust performance compared with existing baselines, particularly under challenging random control and unseen epitope settings. These results highlight LANTERN’s potential for robust TCR-pMHC binding prediction and downstream applications in personalized immunotherapy and vaccine development. For reproducing, our code is available at: https://anonymous.4open.science/r/LANTERN-87D9 .
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Computer Science, New Jersey Institute of Technology, Newark, New Jersey, United States.
ORCID-Profile: Cong Qi
Förderung
- NIGMS NIH HHS
Zitieren
APA
Qi, C., Fang, H., Jiang, S., Hu, T., & Wei, Z. (2026). LANTERN: TCR-peptide binding prediction via large language model representations. PeerJ, 14, e20980. https://doi.org/10.7717/peerj.20980
Vancouver
Qi C, Fang H, Jiang S, Hu T, Wei Z. LANTERN: TCR-peptide binding prediction via large language model representations. PeerJ. 2026;14:e20980. doi:10.7717/peerj.20980
BibTeX
@article{Qi2026,
author = {Qi, Cong and Fang, Hanzhang and Jiang, Siqi and Hu, Tianxing and Wei, Zhi},
title = {LANTERN: TCR-peptide binding prediction via large language model representations},
journal = {PeerJ},
year = {2026},
volume = {14},
pages = {e20980},
doi = {10.7717/peerj.20980}
}
EUVIMED