Scanpy

You can run the Scanpy-MCP server to control Scanpy analysis workflows through natural language requests. This MCP server exposes data I/O, preprocessing, analysis tools, and plotting capabilities, enabling you to build AI-driven pipelines that operate Scanpy from any MCP client.

How to use

Start by installing the package and running the MCP server, then connect your MCP client to the server URL or run it locally via stdio. You will issue commands through your client to load data, preprocess, run clustering or differential expression analyses, and generate plots, all mediated by natural language interactions.

Typical usage patterns include starting the server locally and pointing your MCP client to the provided local URL, or running the server as a stdio process and configuring the client to invoke it directly. This setup supports both on-device and remote workflows, letting you dispatch Scanpy operations without writing raw Python code.

How to install

Prerequisites
- Python 3.8+ installed on your system
- Access to a Python package manager (pip) and a compatible MCP client
- Optional: Node.js environment if your MCP client relies on npm/npx technologies for orchestration

Install the Scanpy-MCP package from PyPI and verify the installation by running the MCP server in test mode.

pip install scanpy-mcp

scanpy-mcp run

Configuration and usage notes

Two primary ways to run Scanpy-MCP are provided below. Use the one that matches your environment and integration needs.

// Local stdio (direct process)
"mcpServers": {
  "scanpy-mcp": {
    "command": "//home/test/bin/scanpy-mcp",
    "args": [
      "run"
    ]
  }
}

Additional configuration examples

If you prefer to expose the MCP server over HTTP for remote clients, start the server with transport over HTTP and expose a port, then point your MCP client to the HTTP URL.

scanpy-mcp run --transport shttp --port 8000

// Client configuration (remote)
"mcpServers": {
  "scanpy-mcp": {
    "url": "http://localhost:8000/mcp"
  }
}

Available tools

io

IO module to read and write scRNA-Seq data.

preprocessing

Preprocessing module for filtering, QC, normalization, scaling, HVG, PCA, neighbors, and related steps.

tool

Tool module for clustering, differential expression, and other analyses.

plotting

Plotting module for violin, heatmap, dotplot, and related visualizations.