PDBe MCP Servers

First-party MCP servers from the Protein Data Bank in Europe (PDBe) that let Claude query structural data through the PDBe REST API, run advanced Solr searches, and optionally traverse the PDBe knowledge graph.


Type	MCP server
Supplier	PDBe (EMBL-EBI)
Availability	GA — `pdbe-mcp-server` on PyPI (v1.1.4, 2026-06-15)
Pricing	Free / OSS (Apache-2.0)
Capabilities	Read-only
Verified	works · 2026-07-20
Security	cleared · 2026-07-20 — provenance matches PDBe EMBL-EBI, Apache-2.0, keyless read-only

How to install

The package ships three servers selected by a --server-type flag. The API and Search servers need no accounts, API keys, or extra infrastructure; the Graph server additionally requires a self-hosted Neo4j instance.

Prerequisite: uvx requires uv installed — pip install uv or the official installer.

Claude Code — direct MCP add over stdio (Claude launches the process itself via uvx; no separate terminal needed):

claude mcp add --transport stdio pdbe-api -- uvx pdbe-mcp-server --server-type pdbe_api_server
claude mcp add --transport stdio pdbe-search -- uvx pdbe-mcp-server --server-type pdbe_search_server

Claude Desktop — add the equivalent stdio entries to claude_desktop_config.json:

{
  "mcpServers": {
    "PDBe API Server": {
      "command": "uvx",
      "args": ["pdbe-mcp-server", "--server-type", "pdbe_api_server"]
    },
    "PDBe Search Server": {
      "command": "uvx",
      "args": ["pdbe-mcp-server", "--server-type", "pdbe_search_server"]
    }
  }
}

Graph server (optional, requires Neo4j) — set NEO4J_URL, NEO4J_USERNAME, and NEO4J_PASSWORD for a running Neo4j instance, then register:
```
claude mcp add --transport stdio pdbe-graph --env NEO4J_URL=bolt://localhost:7687 --env NEO4J_USERNAME=neo4j --env NEO4J_PASSWORD=<your-password> -- uvx pdbe-mcp-server --server-type pdbe_graph_server
```
(Replace <your-password> with your Neo4j password. PDBe does not host a public graph endpoint, so the Graph server is only useful if you stand up your own Neo4j mirror of the PDBe graph.)

To confirm a server boots before registering, run the uvx … command directly and Ctrl-C after it starts — Claude Code/Desktop will launch the process on demand via stdio, so you do not keep it running in a separate terminal.

What it does

Three complementary servers over PDBe resources:

PDBe API Server (pdbe_api_server) — tools auto-generated from the PDBe REST API OpenAPI spec: entries, assemblies, molecules, ligands, publications, and structure validation data.
PDBe Search Server (pdbe_search_server) — get_pdbe_search_schema and run_pdbe_search_query for advanced Solr-based faceted searches across structural data.
PDBe Graph Server (pdbe_graph_server) — pdbe_graph_nodes, pdbe_graph_edges, pdbe_graph_node_relationships, pdbe_graph_example_queries, and pdbe_run_cypher_query for exploring the PDBe knowledge-graph schema and running Cypher queries (against a local Neo4j).

Primary use cases: Retrieve PDBe structure metadata and validation into Claude workflows, run faceted Solr queries over the archive, explore structural relationships via the PDBe graph.

Notes

Claude Science: This resource is offered inside Anthropic’s Claude Science via the Structures & Interactions featured connector. Its inclusion there is an independent signal of quality and trustworthiness for life-science research.

Apache-2.0. The API and Search servers are keyless and run out of the box; the Graph server needs a self-hosted Neo4j (PDBe does not expose a public graph endpoint), so most users should start with the API and Search servers. All servers are read-only. This is a PDBe Europe resource (EMBL-EBI), distinct from the RCSB-focused servers on the PDB MCP Server page — PDBe and RCSB are separate wwPDB partner sites with different APIs, search backends, and graph infrastructure; catalogued separately as one entry per data source.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.