CellVoyager is an open-source project designed to leverage Large Language Models (LLMs) for automated analysis and hypothesis generation in single-cell genomics. It targets researchers and bioinformaticians who need to explore complex biological datasets, providing an interactive agent that can guide analyses, generate hypotheses, and build Jupyter notebooks. The primary benefit is accelerating biological discovery by automating tedious analytical steps and suggesting novel research directions.

How It Works

CellVoyager employs a multi-LLM approach, allowing users to select different models for execution and hypothesis generation. It integrates with popular single-cell data formats like AnnData (.h5ad) and generates interactive Jupyter notebooks that mirror the agent's analysis steps. Users can guide the agent through a GUI or terminal interface, providing biological context, specifying research directions, and offering feedback at various stages, enabling a collaborative and iterative analysis process. The "DeepResearch" feature further enhances hypothesis generation by querying external LLM agents for additional biological background.

Quick Start & Requirements

• Installation: Clone the repository, create and activate a Conda environment using environment.yml.
• Prerequisites: API keys for OpenAI and Anthropic are required, configured via a .env file (OPENAI_API_KEY, ANTHROPIC_API_KEY).
• Running:
  • GUI: streamlit run gui/app.py
  • Terminal: python run_cellvoyager.py --h5ad-path <path_to_h5ad> --paper-path <path_to_summary> --analysis-name <run_name>
• Example Data: A COVID-19 dataset (.h5ad) can be downloaded using curl.
• Links: Demo UI available (uses pre-loaded data).

Highlighted Details

• Interactive GUI and terminal modes allow real-time user feedback and code editing.
• Supports multiple LLM backends for execution and hypothesis generation.
• "DeepResearch" option integrates external LLM agents for enhanced biological context.
• Simultaneously builds Jupyter notebooks for transparency and manual extension.

Maintenance & Community

No specific details on contributors, sponsorships, community channels (like Discord/Slack), or roadmap were found in the provided README.

Licensing & Compatibility

The license type is not explicitly stated in the provided README.

Limitations & Caveats

The demo UI has memory constraints and uses a pre-loaded dataset. API keys for specific LLM providers are mandatory for full functionality. The project appears to be research-oriented, and its stability or suitability for production environments is not detailed.