PubTator MCP Server

What is PubTator MCP Server?

PubTator MCP Server is a biomedical literature annotation and relationship mining server based on PubTator3, providing access through the Model Context Protocol (MCP). It allows AI models to programmatically search scientific literature, obtain annotation information, and analyze entity relationships.

How to use PubTator MCP Server?

Install via Smithery or manually (clone repo, pip install -r requirements.txt) with Python 3.10+ and FastMCP. Run the server directly (python pubtator_server.py) or via Docker. Configure transport (TCP default, stdio available) and host/port via environment variables. Use MCP client tools (e.g., Claude Desktop, Cursor, CLine) to invoke functions such as export_publications, find_entity_id, find_related_entities, search_pubtator, and batch_export_from_search.

Key features of PubTator MCP Server

• Literature Annotation Export in multiple formats (pubtator, biocxml, biocjson)
• Entity ID Lookup for biological concepts via free text
• Relationship Mining to discover biomedical relationships between entities
• Literature Search by keywords and entity IDs
• Batch Processing of annotation export from search results

Use cases of PubTator MCP Server

• Automating extraction of annotated biomedical entities from PubMed articles
• Querying standardized identifiers for genes, diseases, chemicals, and species
• Discovering relationships like treatments, causes, or interactions between entities
• Performing large-scale batch annotation export for research datasets
• Integrating biomedical literature mining into AI assistant workflows

FAQ from PubTator MCP Server

What are the runtime requirements and dependencies?

Python 3.10+ and the FastMCP library are required. Docker is also supported for containerized deployment.

What transport protocols does the server support?

Both stdio and TCP transports are supported. The default is TCP on host 0.0.0.0 port 8080, configurable via MCP_TRANSPORT, MCP_HOST, and MCP_PORT environment variables.

What are the usage limitations?

API request rate limit is maximum 3 requests per second. When batch exporting, use a reasonable batch_size to avoid timeout. For relationship queries, entity IDs must start with "@" (e.g., @DISEASE_COVID-19).

Where does the data come from?

The server integrates with PubTator3, a biomedical literature annotation system. All data originates from PubTator’s API.

Is there a disclaimer or license?

This project is licensed under MIT. It is intended for research purposes only; users must comply with PubTator’s terms of service.