Autonomous Bioinformatics Agent

An AI-powered pipeline that takes a plain-English research question, discovers relevant NCBI GEO datasets, downloads and preprocesses them, identifies top expressed genes, and produces a Markdown report with Gemini-generated biological interpretation.

🚀 Features

• Intelligent Planning: Gemini decomposes your prompt into a logical task sequence.
• Smart Discovery: Uses Gemini to refine natural language questions into precise NCBI GEO search queries.
• Automated Workflow:
  • Fetches and decompresses GEO series matrix files.
  • Cleans and normalizes expression data (log2 transformation).
  • Ranks genes by expression levels across samples.
• Biological Interpretation: Gemini explains the scientific significance of the top genes found.
• Structured Reporting: Generates comprehensive Markdown reports in the reports/ directory.

🛠️ Installation

1. Clone the repository:

   git clone https://github.com/yourusername/autonomous-bioinformatics-agent.git
   cd autonomous-bioinformatics-agent

2. Set up a virtual environment:

   python3 -m venv .venv
   source .venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Set your Gemini API Key: The agent requires a Google Gemini API key. You can get one from Google AI Studio.

   export GEMINI_API_KEY="your_api_key_here"

📈 Usage

You can run the agent by passing your research question as a command-line argument:

python main.py "Which genes are most highly expressed in breast cancer GSE2034?"

Or run it interactively by just calling the script:

python main.py

🏗️ Project Structure

.
├── agents/             # AI Agents for planning, discovery, and interpretation
├── tools/              # Core bioinformatics tools for analysis and reporting
├── data/               # Local storage for raw and processed datasets (ignored by git)
├── reports/            # Generated Markdown reports (ignored by git)
├── tests/              # Pytest suite for code verification
├── config.py           # Global settings and model configuration
├── main.py             # Main entry point
└── requirements.txt    # Project dependencies

⚙️ Configuration

Defaults can be adjusted in config.py:

• GEMINI_MODEL: The model version used (defaults to models/gemini-flash-latest).
• TOP_GENE_COUNT: Number of top-ranked genes to include in reports.
• MAX_DOWNLOAD_RETRIES: Number of attempts for fetching remote datasets.

🧪 Testing

Run the test suite to ensure everything is configured correctly:

pytest tests/

⚖️ License

This project is open-source. See the LICENSE for details (if available).