scAgent: Universal Single-Cell Annotation via a LLM Agent
This addresses the challenge of generalizing cell annotation across tissues and discovering novel cell types, which is incremental by applying LLMs to a known bottleneck in single-cell analysis.
The paper tackles the problem of universal cell type annotation across diverse tissues, proposing scAgent, an LLM-based framework that achieves superior performance in general annotation, novel cell discovery, and extensibility to novel cell types, as demonstrated in experiments on 160 cell types and 35 tissues.
Cell type annotation is critical for understanding cellular heterogeneity. Based on single-cell RNA-seq data and deep learning models, good progress has been made in annotating a fixed number of cell types within a specific tissue. However, universal cell annotation, which can generalize across tissues, discover novel cell types, and extend to novel cell types, remains less explored. To fill this gap, this paper proposes scAgent, a universal cell annotation framework based on Large Language Models (LLMs). scAgent can identify cell types and discover novel cell types in diverse tissues; furthermore, it is data efficient to learn novel cell types. Experimental studies in 160 cell types and 35 tissues demonstrate the superior performance of scAgent in general cell-type annotation, novel cell discovery, and extensibility to novel cell type.