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

Data-driven methods, including unsupervised and supervised learning, have become essential tools for analyzing single-cell RNA sequencing (scRNA-seq) data. However, these approaches cannot incorporate research context into the analysis, limiting their ability to provide in-depth, contextual insights to researchers. We introduce scChat, a pioneering AI assistant designed for contextualized scRNA-seq analysis. Unlike existing tools that focus on standard tasks like cell annotation, scChat offers advanced capabilities, including research context-based analysis of experimental outcomes, detailed validation of research hypotheses, identification of potential explanations when hypotheses fail, and suggestions for next-step experimental design. Powered by a large language model (LLM), scChat integrates function calls for quantitative tasks, employs retrieval-augmented generation to reduce hallucination, and uses an LLM-powered search engine to seamlessly incorporate current literature into the analytical conversation.