概要
What is Pubtator MCP Server?
An async Python server that integrates with the PubTator3 API to provide biomedical text‑mining tools via the MCP agent protocol. It supports entity lookup, literature search, and article text extraction for researchers and developers.
How to use Pubtator MCP Server?
Install with pip install mcp-server-pubtator3, then run mcp-server-pubtator3. The server exposes four MCP tools (find_entity, search_pubtator, get_paper_text, find_related_entities) that can be called from any MCP‑compatible platform.
Key features of Pubtator MCP Server
- Entity autocomplete for genes, diseases, chemicals, and variants
- Literature search by keywords, IDs, or entity relations
- Article retrieval from PubMed/PMC in plain text
- Find entities related to a given identifier with filters
- Async, non‑blocking operation via
aiohttp
Use cases of Pubtator MCP Server
- Identify biomedical concept identifiers from free‑text queries
- Search for articles about a specific gene‑disease interaction
- Extract full‑text content from PubMed or PMC articles
- Discover related entities (e.g., chemicals associated with a disease)
- Integrate biomedical knowledge into agent‑based analysis pipelines
FAQ from Pubtator MCP Server
What Python version is required?
Python 3.10 or higher is required.
Do I need an API key or authentication?
No authentication is mentioned; the server directly queries the public PubTator3 API.
What is the maximum number of results per query?
Each tool’s limit parameter has a default of 10 and a maximum of 50.
What does the get_paper_text tool return?
It returns the plain text content of the article if available.
How do I access the available tools?
The tools are listed in the list_tools endpoint and can be invoked over the MCP protocol.
