ChatCellAnno: Multimodal AI-Powered Single-Cell Annotation Assistant

ChatCellAnno is a desktop application for single-cell annotation that combines marker genes, optional expression matrices, enrichment evidence, and visual context for LLM-assisted cluster labeling.

中文文档

Features

• Multimodal context: marker genes, optional UMAP image context, and optional expression matrix.
• Enrichment-assisted prompting: integrates GO/KEGG and related knowledge to reduce annotation ambiguity.
• Flexible interaction modes: built-in browser flow plus configurable API settings.
• Structured parsing: turns LLM responses into cluster annotations.
• Code export: generates Scanpy (Python) and Seurat (R) annotation snippets.
• Modular GUI: PySide6-based layout with separate core logic and UI modules.

Project Structure

• chatcellanno/core.py: workflow orchestration (generate and parse).
• chatcellanno/extractor.py: marker extraction from CSV/TSV formats.
• chatcellanno/prompt.py: prompt construction.
• chatcellanno/parser.py: response parsing and code generation.
• chatcellanno/enrichment.py: enrichment analysis integration.
• chatcellanno/gui/: main window, UI blocks, and worker threads.
• gui.py: desktop app entry point.

Installation

Requirements

• Python 3.10+

Quick Start

conda create -n chatcellanno python=3.10
conda activate chatcellanno
pip install pandas pyperclip gseapy biopython PySide6 PySide6-WebEngine
python gui.py

Build (Windows)

./build.ps1

The executable is generated under dist/.

Typical Workflow

1. Load marker file (.csv or .tsv).
2. Optionally add enrichment and image/matrix context.
3. Generate a prompt file in Step 2, then either drag it to the built-in browser chat (manual mode) or send it automatically with API mode.
4. Parse the LLM response and export code/report in Step 4 (Parsing & Report).

Development

• Run tests from tests/ to verify parser and enrichment behavior.
• Keep large runtime data (database/, genome/, results/) out of source control.

License

MIT. See LICENSE.