CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis

TL;DR

CellMaster is an AI-powered tool that uses large language models to accurately annotate cell types in single-cell RNA-seq data, especially excelling in identifying rare and novel cell states without prior marker databases.

Contribution

It introduces a zero-shot, LLM-based annotation system that improves accuracy over existing methods and supports human-in-the-loop refinement for complex cell type identification.

Findings

Achieved 7.1% higher accuracy than baselines in automatic mode.

Enhanced accuracy by 18.6% with human-in-the-loop refinement.

Performed well in identifying rare and novel cell states.

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 application are available at \href{https://github.com/AnonymousGym/CellMaster}{https://github.com/AnonymousGym/CellMaster}.