Uniprot

UniProt MCP Server - Access comprehensive protein data via Model Context Protocol. Search, retrieve, and analyze UniProt entries with full-text search, ID mapping, and rich metadata. Supports stdio & HTTP transports for Claude Desktop and custom integrations.

Installation

Run this command in your terminal to add the MCP server to Claude Code.

Run in terminal:

View docs

Command

claude mcp add --transport stdio josefdc-uniprot-mcp uvx uniprot-mcp \
  --env MCP_HTTP_HOST="HTTP server bind address (optional if using stdio only)" \
  --env MCP_HTTP_PORT="HTTP server port (optional if using stdio only)" \
  --env MCP_HTTP_RELOAD="Enable auto-reload for HTTP server (1 or true to enable)" \
  --env UNIPROT_LOG_LEVEL="Logging level (debug, info, warning, error) (optional)" \
  --env MCP_HTTP_LOG_LEVEL="Uvicorn log level for HTTP server (optional)" \
  --env UNIPROT_LOG_FORMAT="Log format: plain or json (optional)" \
  --env UNIPROT_ENABLE_FIELDS="Request minimal field subsets to reduce payload size (optional)" \
  --env UNIPROT_MAX_CONCURRENCY="Max concurrent UniProt API requests (optional)"

How to use

UniProt MCP Server provides a typed, MCP-compliant interface to access UniProtKB data. It exposes a set of MCP primitives (Resources, Tools, and Prompts) that let you fetch complete protein entries with sequences and annotations, perform advanced searches, map identifiers across databases, and retrieve historical entry versions. You can interact with the server via the stdio-based MCP interface for local development or the HTTP endpoint for remote deployments. Typical usage involves querying for a protein entry, extracting its sequence and GO annotations, or performing an ID mapping across databases, all through well-typed responses that are friendly to LLM agents.

How to install

Prerequisites:

Python 3.11 or 3.12
pip (or uv for development)

Install the UniProt MCP package:

pip install uniprot-mcp

Run the local stdio server (default behavior):

uniprot-mcp

Run the HTTP server (remote deployment):

uniprot-mcp-http --host 0.0.0.0 --port 8000

Optionally, run via the MCP development workflow (uv for development):

uvx uniprot-mcp

Additional notes

Environment variables can customize behavior and performance:

UNIPROT_ENABLE_FIELDS, UNIPROT_LOG_LEVEL, UNIPROT_LOG_FORMAT
UNIPROT_MAX_CONCURRENCY to limit parallel API calls
MCP_HTTP_HOST / MCP_HTTP_PORT / MCP_HTTP_RELOAD to control the HTTP server If you run HTTP, the MCP endpoint is at http://<host>:<port>/mcp and a health check is available at /healthz with metrics at /metrics (Prometheus format). For development, you can test commands with MCP Inspector:

npx @modelcontextprotocol/inspector uniprot-mcp

Related MCP Servers

npcpy

1.2k

The python library for research and development in NLP, multimodal LLMs, Agents, ML, Knowledge Graphs, and more.

sample-agentic-ai-demos

100

Collection of examples of how to use Model Context Protocol with AWS.

mcp-python-interpreter

MCP Python Interpreter: run python code. Python-mcp-server, mcp-python-server, Code Executor

mcp-gateway

MCP Gateway and Registry

mcpx-py

Python client library for https://mcp.run - call portable & secure tools for your AI Agents and Apps

muxi

An extensible AI agents framework