scmcphub’s MCP servers

scmcphub provides a collection of MCP servers for single-cell analysis. These servers enable natural language interaction with various single-cell analysis tools. You can install any of the specialized MCP servers based on your needs.

  • scanpy-mcp: A comprehensive MCP server for basic single-cell analysis operations, including data preprocessing, dimensionality reduction, clustering, and visualization.

  • cellrank-mcp: An MCP server specialized in cell fate analysis and trajectory inference. It enables natural language interaction with CellRank’s capabilities for analyzing cell state transitions, fate probabilities, and developmental trajectories.

  • liana-mcp: A dedicated MCP server for cell-cell communication analysis. It provides natural language interfaces to LIANA’s methods for inferring and analyzing cell-cell communication from single-cell data.

  • decoupler-mcp: An MCP server focused on pathway and gene set analysis. It enables natural language interaction with decoupler’s tools for pathway activity inference and gene set analysis in single-cell data.

  • infercnv-mcp: An MCP server focused on CNV analysis. It enables natural language interaction with infercnvpy’s tools for CNV analysis in single-cell data.

  • scmcp: A unified MCP server that integrates all single-cell analysis capabilities. It provides a comprehensive natural language interface for the entire single-cell analysis workflow, from data preprocessing to advanced analysis, combining the functionalities of all specialized MCP servers into one powerful tool.

Installation

Install the MCP servers using pip:

pip install scmcp
pip install scanpy-mcp
pip install cellrank-mcp
pip install liana-mcp
pip install decoupler-mcp
pip install infercnv-mcp

For faster installation, use uv:

uv pip install scmcp scanpy-mcp cellrank-mcp liana-mcp decoupler-mcp infercnv-mcp

Verify Installation

To verify that the MCP servers are installed correctly, you can run the following commands in your terminal:

# Verify scmcp installation
scmcp run

# Verify scanpy-mcp installation
scanpy-mcp run

# Verify cellrank-mcp installation
cellrank-mcp run

# Verify liana-mcp installation
liana-mcp run

# Verify decoupler-mcp installation
decoupler-mcp run

Verify infercnv-mcp installation

infercnv-mcp run


Each command will start the respective MCP server, and you should see output indicating that the server is running.  

## Finding MCP Server Executables

If you need to locate the MCP server executables on your system:

```bash
# On Linux/Mac
which scmcp
which scanpy-mcp
which cellrank-mcp
which liana-mcp
which decoupler-mcp
which infercnv-mcp

# On Windows (PowerShell)
Get-Command scmcp
Get-Command scanpy-mcp
Get-Command cellrank-mcp
Get-Command liana-mcp
Get-Command decoupler-mcp
Get-Command infercnv-mcp

# On Windows (Command Prompt)
where scmcp
where scanpy-mcp
where cellrank-mcp
where liana-mcp
where decoupler-mcp
where infercnv-mcp