scChat: A Large Language Model-Powered Co-Pilot for Contextualized Single-Cell RNA Sequencing Analysis

Abstract

Single-cell RNA sequencing (scRNA-seq) has transformed biomedical research by enabling transcriptomic analysis at single-cell resolution. Yet, existing computational approaches remain primarily data-driven and lack the ability to integrate research context, limiting their interpretability and impact on hypothesis generation or experimental planning. We present scChat, a large language model (LLM)-powered co-pilot for contextualized scRNA-seq analysis. Unlike conventional pipelines restricted to tasks such as cell type annotation or enrichment analysis, scChat has an interactive, reasoning-based framework. It combines quantitative algorithms with retrieval-augmented generation and a multi-agent architecture to support hypothesis validation, mechanistic interpretation, and next-step experimental design. Through showcase and benchmarking studies, we demonstrate that scChat not only achieves high accuracy in cell type annotation but also provides biologically grounded explanations and contextual insights.