PDBe s
PDBe MCP Servers integrate Protein Data Bank Europe resources with LLMs via Model Context Protocol. Provides seamless access to protein structure data through API tools and graph database schema assistance for intelligent Cypher query generation, bridging structural biology and AI research.
claude mcp add --transport stdio pdbeurope-pdbe-mcp-servers uvx pdbe-mcp-server --server-type pdbe_api_server
How to use
PDBe MCP Servers expose three tools that map to PDBe APIs: the PDBe API Server provides core REST API access to structural data, the PDBe Graph Server lets you query the PDBe Graph Database to explore relationships and interactions, and the PDBe Search Server enables advanced Solr-based search across PDBe structures. These tools are exposed as MCP endpoints that can be invoked from Claude Desktop or other MCP-enabled clients. You can interact with the servers using the uvx command (for PyPI installations) or via the local development workflow, selecting the appropriate server type when starting the MCP server. Typical usage includes searching for structures by UniProt accession, exploring binding sites and molecular interactions, and running complex structure searches with filters such as resolution or experimental method.
How to install
Prerequisites:
- Python 3.10+
- uv package manager (from the uv project)
Option A - Quick Start (PyPI installation):
- Install and run directly from PyPI:
uvx pdbe-mcp-server
This uses the PDBe MCP Server package from PyPI and starts the server with default settings. The server endpoints are accessible via the configured transport (default SSE at http://localhost:8000/sse).
Option B - Local Development Installation:
- Clone the repository:
git clone https://github.com/PDBeurope/PDBe-MCP-Servers.git
cd PDBe-MCP-Servers
- Create a virtual environment (uv venv):
uv venv
- Install in editable/development mode:
uv pip install .
Note: For development, you can install with editable mode and customize the server packages as needed.
General usage for starting servers (two options):
- PyPI installation:
uvx pdbe-mcp-server --server-type pdbe_api_server --transport sse
uvx pdbe-mcp-server --server-type pdbe_graph_server --transport sse
uvx pdbe-mcp-server --server-type pdbe_search_server --transport sse
- Local development (from repository):
uv run pdbe-mcp-server --server-type pdbe_api_server --transport sse
uv run pdbe-mcp-server --server-type pdbe_graph_server --transport sse
uv run pdbe-mcp-server --server-type pdbe_search_server --transport sse
The server defaults to http://localhost:8000/sse when using the SSE transport.
Additional notes
Tips and notes:
- Ensure Python 3.10+ is installed and available in PATH.
- If you install via PyPI, uvx is included with uv; make sure uvx is accessible from your shell.
- For local development, point the server to your PDBe-MCP-Servers directory when using the conditional uv run command.
- Claude Desktop integration requires adding three MCP entries: PDBe API Server, PDBe Graph Server, and PDBe Search Server, each using command uvx with the appropriate --server-type. The exact naming in Claude can be adjusted to your preference.
- The PDBe Search Server supports get_search_schema and run_search_query operations; refer to the README for example query formats and field names (e.g., pdb_id, experimental_method, release_date, resolution).
- If you encounter port or transport issues, verify that the server is reachable at the expected URL (default http://localhost:8000/sse) and that your firewall allows local connections.
Related MCP Servers
mcp-vegalite
MCP server from isaacwasserman/mcp-vegalite-server
github-chat
A Model Context Protocol (MCP) for analyzing and querying GitHub repositories using the GitHub Chat API.
nautex
MCP server for guiding Coding Agents via end-to-end requirements to implementation plan pipeline
pagerduty
PagerDuty's official local MCP (Model Context Protocol) server which provides tools to interact with your PagerDuty account directly from your MCP-enabled client.
futu-stock
mcp server for futuniuniu stock
mcp -boilerplate
Boilerplate using one of the 'better' ways to build MCP Servers. Written using FastMCP