![[2602.13346] CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.13346] CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis
Summary
CellMaster introduces an AI-driven approach for zero-shot cell-type annotation in single-cell RNA sequencing, improving accuracy significantly over existing methods.
Why It Matters
This research addresses a critical bottleneck in genomics by enhancing the accuracy of cell-type identification, which is vital for understanding complex biological systems and diseases. The use of AI in this context could revolutionize how researchers annotate and analyze single-cell data, leading to more precise insights in biomedical research.
Key Takeaways
- CellMaster improves cell-type annotation accuracy by 7.1% over existing tools.
- With human input, accuracy gains increase to 18.6%, especially for rare cell types.
- The system utilizes LLM-encoded knowledge for on-the-fly annotation without pre-training.
- CellMaster demonstrates strength in identifying novel and rare cell states.
- Source code and web application are available for further research and application.
Quantitative Biology > Genomics arXiv:2602.13346 (q-bio) [Submitted on 12 Feb 2026] Title:CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis Authors:Zhen Wang, Yiming Gao, Jieyuan Liu, Enze Ma, Jefferson Chen, Mark Antkowiak, Mengzhou Hu, JungHo Kong, Dexter Pratt, Zhiting Hu, Wei Wang, Trey Ideker, Eric P. Xing View a PDF of the paper titled CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis, by Zhen Wang and 12 other authors View PDF HTML (experimental) Abstract:Single-cell RNA-seq (scRNA-seq) enables atlas-scale profiling of complex tissues, revealing rare lineages and transient states. Yet, assigning biologically valid cell identities remains a bottleneck because markers are tissue- and state-dependent, and novel states lack references. We present CellMaster, an AI agent that mimics expert practice for zero-shot cell-type annotation. Unlike existing automated tools, CellMaster leverages LLM-encoded knowledge (e.g., GPT-4o) to perform on-the-fly annotation with interpretable rationales, without pre-training or fixed marker databases. Across 9 datasets spanning 8 tissues, CellMaster improved accuracy by 7.1% over best-performing baselines (including CellTypist and scTab) in automatic mode. With human-in-the-loop refinement, this advantage increased to 18.6%, with a 22.1% gain on subtype populations. The system demonstrates particular strength in rare and novel cell states where baselines often fail. Source code and the web applicat...
