EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Benchmarking large language models for cell typing in single-cell RNA-Seq

Tianxiang Xiao, Dezhi Hua, Yanan Wang iD, Xuemei Lu iD, Chao Zhang iD

Briefings in Bioinformatics · 01.11.2025 · 26 (6)

DOI: 10.1093/bib/bbaf677 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Briefings in Bioinformatics, online erschienen am 15.12.2025. Verfasst von Tianxiang Xiao und 4 weiteren Autor:innen. Beteiligte Einrichtung: State Key Laboratory of Genetic Evolution & Animal Models, Kunming Institute of Zoology, Chinese Academy of Sciences, No. 17 Longxin Road, Panlong Di…. Hauptthemen laut MEDLINE: Single-Cell Analysis, RNA-Seq, Sequence Analysis, RNA. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Pioneer Hundred Talents Program of the Chinese Academy of Sciences and the Yunnan Talent Support Plan, National Natural Science Foundation of China, Yunnan Basic Research Special Project-Key Project.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Large language models (LLMs) show significant potential for cell type annotation in single-cell RNA sequencing (scRNA-seq), but a systematic framework for their application and a comprehensive performance comparison are lacking. Here, we systematically benchmarked seven leading LLM models and three traditional bioinformatics tools across 34 diverse human and mouse datasets. We establish that using marker genes selected by statistical significance and ranked by log₂ fold change optimizes annotation accuracy. Our benchmark reveals that LLMs profoundly outperform traditional methods, particularly in resolving fine-grained cell subtypes. A top tier of models, including Kimi-k2, GPT-5, Claude-4.1, and Grok-4, consistently delivered the highest accuracy. To harness their collective strength, we developed an elite ensemble strategy that achieves state-of-the-art performance. We encapsulated these findings into DeepCellSeek, an open-source R package and interactive web platform, to provide a validated, high-performance solution. This work provides a practical roadmap for leveraging LLMs in single-cell research and paves the way for their evolution into powerful discovery engines.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Single-Cell AnalysisRNA-SeqSequence Analysis, RNAHumansMiceBenchmarkingAnimalsComputational BiologySoftwareLarge Language ModelsSingle-Cell Gene Expression Analysis

Stichwörter der Autor:innen

DeepCellSeekbenchmarkcell type annotationensemble strategylarge language modelsingle-cell RNA sequencing

Autor:innen und Institutionen

  • State Key Laboratory of Genetic Evolution & Animal Models, Kunming Institute of Zoology, Chinese Academy of Sciences, No. 17 Longxin Road, Panlong District, Kunming, Yunnan 650201, China.
  • College of Life Sciences, University of Chinese Academy of Sciences, No. 19(A) Yuquan Road, Shijingshan District, Beijing 100049, China.
  • Yunnan Key Laboratory of Biodiversity Information, Kunming Institute of Zoology, Chinese Academy of Sciences, No. 17 Longxin Road, Panlong District, Kunming, Yunnan 650201, China.
  • National Resource Center for Non-Human Primates, Kunming Primate Research Center, and National Research Facility for Phenotypic and Genetic Analysis of Model Animals (Primate Facility), Kunming Institute of Zoology, Chi…
  • Biodiversity Data Center of Kunming Institute of Zoology, Chinese Academy of Sciences, No. 17 Longxin Road, Panlong District, Kunming, Yunnan 650201, China.

ORCID-Profile: Yanan Wang, Xuemei Lu, Chao Zhang

Förderung

  • Pioneer Hundred Talents Program of the Chinese Academy of Sciences and the Yunnan Talent Support Plan
  • National Natural Science Foundation of China
  • Yunnan Basic Research Special Project-Key Project
  • Yunnan Revitalization Talent Support Program Top team
  • Yunnan Basic Research Special Project—Key Project

Zitieren

APA

Xiao, T., Hua, D., Wang, Y., Lu, X., & Zhang, C. (2025). Benchmarking large language models for cell typing in single-cell RNA-Seq. Briefings in Bioinformatics, 26(6). https://doi.org/10.1093/bib/bbaf677

Vancouver

Xiao T, Hua D, Wang Y, Lu X, Zhang C. Benchmarking large language models for cell typing in single-cell RNA-Seq. Briefings in Bioinformatics. 2025;26(6). doi:10.1093/bib/bbaf677

BibTeX
@article{Xiao2025,
  author = {Xiao, Tianxiang and Hua, Dezhi and Wang, Yanan and Lu, Xuemei and Zhang, Chao},
  title = {Benchmarking large language models for cell typing in single-cell RNA-Seq},
  journal = {Briefings in Bioinformatics},
  year = {2025},
  volume = {26},
  number = {6},
  doi = {10.1093/bib/bbaf677}
}
Kostenlos starten →