Powerful Model Context Protocol (MCP) server for therapeutic discovery and drug repurposing using the Open Targets Platform GraphQL API.
Open Targets MCP Server 🧬
A powerful, workflow-oriented Model Context Protocol (MCP) server for therapeutic discovery and drug repurposing. This server provides a seamless bridge between agentic AI platforms and the Open Targets Platform, leveraging its comprehensive GraphQL API.
🚀 Features
The server implements a suite of specialized tools for automated drug discovery:
- Entity Resolution: Resolve common names (e.g., 'BRCA2', 'aspirin', 'asthma') to standardized identifiers (Ensembl, ChEMBL, EFO).
- Target Prioritization: Retrieve top associated targets for any disease, ranked by multi-faceted association scores.
- Safety Assessments: Quickly pull safety liabilities, toxicities, and risk information for gene/protein targets.
- Pharmacovigilance: Access statistical FDA FAERS data (adverse events) for known drugs using log-likelihood ratios.
- Druggability Analysis: View comprehensive tractability summaries for targets across multiple modalities (Small Molecule, Antibody, etc.).
- Disease Phenotyping: Map diseases to their clinical signs and symptoms using the Human Phenotype Ontology (HPO).
- Cancer Profiling: Retrieve cancer hallmarks (COSMIC) and gene essentiality scores from the Cancer Dependency Map (DepMap).
- Animal Models: Access significant phenotypes observed in biological mouse models for human targets.
🛠 Tech Stack
- Core: Python 3.10+
- MCP Framework: FastMCP
- API Interface: Open Targets GraphQL API (v4)
- Runtime: HTTP/Streamable transport for modern agentic platforms.
📦 Installation & Setup
-
Clone the repository:
git clone <repository-url> cd OpentargetMCP -
Create and activate a virtual environment:
python3 -m venv .venv source .venv/bin/activate -
Install dependencies:
pip install -r requirements.txt -
Run the server:
python server.py
🤖 Usage with AI Agents
To use this with an MCP-compatible agent (like Claude Desktop or Cursor), add the following to your configuration:
{
"mcpServers": {
"opentarget": {
"command": "python",
"args": ["/path/to/OpentargetMCP/server.py"]
}
}
}
📜 Example Questions
- "Find repurposing candidates for Imatinib and identify its top associated diseases."
- "What are the top therapeutic targets for Alzheimer's and their safety profiles?"
- "Check the cancer essentiality of KRAS across various tissues."
- "What are the clinical signs of Parkinson's Disease according to HPO?"
📄 License
MIT License