opentargets

The Open Targets MCP Server exposes the Open Targets Platform GraphQL API through a suite of MCP tools, enabling researchers and developers to perform target, disease, drug, and evidence analyses via Claude Desktop, web MCP clients, or custom integrations. It provides core capabilities like Target Analysis (gene/protein queries, expression data, pathways, and phenotypes), Disease Analysis (disease queries and associated targets), Drug Discovery (drug queries with safety and indications data), Evidence Mining (target-disease evidence with scoring), Variant Analysis (GWAS and pharmacogenomics data), Study Exploration (GWAS studies and fine-mapped loci), Smart Search (entity resolution and ID mapping), and Cross-Entity Workflows that chain multiple data calls for prioritization. You can also run raw GraphQL operations when needed via the built-in tools. To use the server, start it with your preferred transport (stdio by default) and connect with MCP clients or Claude Desktop. The CLI helpers allow listing tools, getting the version, and adjusting rate limits or logging for troubleshooting.

Prerequisites:

• Python 3.10+ with pip
• Git
• uv (Python Virtual Environment) for local setup (optional but recommended)

Option A: Quick one-shot install via uvx (no local install)

1. Install the server from the GitHub package directly:

uvx --from git+https://github.com/nickzren/opentargets-mcp opentargets-mcp

Option B: Claude Desktop (MCPM) workflow

1. Install the MCP Package Manager and the server:

pip install mcpm
mcpm install opentargets

Option C: Local install for development or self-hosting

1. Clone the repository and install dependencies locally:

git clone https://github.com/nickzren/opentargets-mcp
cd opentargets-mcp
pip install uv

2. Sync dependencies and run the server:

uv sync
uv run python -m opentargets_mcp.server

Option D: Docker (compose)

1. Clone and build with Docker Compose:

git clone https://github.com/nickzren/opentargets-mcp
cd opentargets-mcp
docker-compose up -d --build

Note: The default transport for Docker deployments is http.

Configuration and environment variables are described in the README under Configuration. You can override the API URL, ports, and transport mode as needed.

Tips and common issues:

• Default Open Targets API URL is https://api.platform.opentargets.org/api/v4/graphql; override with OPEN_TARGETS_API_URL if you use a different endpoint or a local/global test instance.
• Transport mode can be stdio, SSE, or HTTP. Use stdio for Claude Desktop or local scripting; use SSE/HTTP for web clients.
• The server enforces structured parameter handling, typed settings, and safety around entity resolution to reduce ambiguity.
• When running locally, set FASTMCP_SERVER_HOST to 127.0.0.1 to limit exposure during development.
• If you hit rate limits, configure OPEN_TARGETS_RATE_LIMIT_RPS and OPEN_TARGETS_RATE_LIMIT_BURST or disable rate limiting for internal testing.
• Use the CLI helpers (opentargets-mcp --list-tools, --version) to discover available tools and verify the server version.