Germline dependence in protein language model retrieval of adaptive immune receptors
Briefings in Bioinformatics · 02.11.2026 · 27 (6)
DOI: 10.1093/bib/bbag614 ↗Kurz gefasst
Fachartikel in der Zeitschrift Briefings in Bioinformatics, online erschienen am 07.11.2026. Verfasst von Suyue Wang und 3 weiteren Autor:innen. Beteiligte Einrichtung: Laboratory of Genome Informatics, Research Institute for Microbial Diseases, The University of Osaka, 3-1 Yamadaoka, Suita 565-0871, Japan. Hauptthemen laut MEDLINE: Receptors, Antigen, T-Cell, Receptors, Immunologic, Adaptive Immunity, Receptors, Antigen, B-Cell. Die Arbeit nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: JST SPRING, Innovative Drug Discovery and Life Science Research, JSPS KAKENHI.
Abstract (Originalsprache)
Protein language models (PLMs) are increasingly used to represent adaptive immune receptors, yet their advantage over classical sequence comparison remains unclear. We benchmarked general and receptor-specific PLMs against classical sequence-comparison methods for antigen-specific nearest-neighbor retrieval across four B- and T-cell receptor databases. Using CDR3 sequences alone, classical methods consistently outperformed general PLMs: BLOSUM62 and Levenshtein exceeded ESM2 variants by 7–12 percentage points in retrieval accuracy. Among receptor-specific models, only SCEPTR and, on one database, TCR2vec matched these baselines, although both additionally encoded V-gene information. AntiBERTa2 and IgBert did not close the performance gap. Using full variable-domain sequences, alignment and PLM performance converged. Germline-reversion, region-masking, and V-gene control analyses indicated that this convergence largely reflected conserved germline-derived context, particularly V-gene identity, with limited evidence for improved extraction of antigen-specific information from the CDRs. Germline dependence also extended to SCEPTR and TCR2vec. Heavily mutated HIV-1 antibodies provided a setting-specific exception to this pattern. Random train-test splits yielded retrieval accuracy 15–28 percentage points higher than clone-aware splits, consistent with optimistic performance estimates due to clonal leakage. These findings emphasize the need for strong sequence-comparison baselines and explicit germline controls when evaluating immune-receptor representations and support clone-aware splitting as a minimum benchmark standard.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Laboratory of Genome Informatics, Research Institute for Microbial Diseases, The University of Osaka, 3-1 Yamadaoka, Suita 565-0871, Japan.
- MyImmune Corporation, 251 Little Falls Drive, Wilmington, DE 19808, United States.
ORCID-Profile: Daron M Standley
Förderung
- JST SPRING
- Innovative Drug Discovery and Life Science Research
- JSPS KAKENHI
Zitieren
APA
Wang, S., Sengupta, A., Li, S., & Standley, D. M. (2026). Germline dependence in protein language model retrieval of adaptive immune receptors. Briefings in Bioinformatics, 27(6). https://doi.org/10.1093/bib/bbag614
Vancouver
Wang S, Sengupta A, Li S, Standley DM. Germline dependence in protein language model retrieval of adaptive immune receptors. Briefings in Bioinformatics. 2026;27(6). doi:10.1093/bib/bbag614
BibTeX
@article{Wang2026,
author = {Wang, Suyue and Sengupta, Ayan and Li, Songling and Standley, Daron M},
title = {Germline dependence in protein language model retrieval of adaptive immune receptors},
journal = {Briefings in Bioinformatics},
year = {2026},
volume = {27},
number = {6},
doi = {10.1093/bib/bbag614}
}
EUVIMED