mcp-biomodelling s

This MCP server suite provides Python-based wrappers around three mechanistic modelling tools: MaBoSS for Boolean/stochastic network simulations, NeKo for network-based cell modelling, and PhysiCell with a settings wrapper for configuring and running PhysiCell simulations. Each tool is exposed as an MCP server via its server.py entrypoint and communicates over standard input/output. To use these tools, configure the MCP in your editor (e.g., VS Code Copilot Chat MCP integration) so that each server is launched through a single Conda-managed Python interpreter. Once configured, you will see the available tools under the MCP /tools interface, with commands corresponding to each server (e.g., maboss, neko, physicell) and their required parameters. You can invoke a tool by name and provide the modelling arguments as you would in the upstream tool’s workflow, and the MCP layer will translate those calls into the underlying Python processes. The servers are designed to share a common Python environment to keep dependencies aligned across MaBoSS, NeKo, and PhysiCell.

Prerequisites:

• A Python-capable environment (preferably via Conda) and Git installed on your system.
• Basic familiarity with the downstream tools (MaBoSS, NeKo, PhysiCell) and their dependencies as described in their respective READMEs.

Installation steps:

1. Create and activate a Conda environment to host all servers:

conda create -n mcp_modelling python=3.11
conda activate mcp_modelling

2. Clone this repository (or ensure you have the mcp-biomodelling-servers folder structure locally):

git clone <repo-url>
cd mcp-biomodelling-servers

3. Install tool-specific dependencies as described in each subfolder README (MaBoSS, NeKo, PhysiCell). This typically involves following upstream installation instructions and ensuring the installed packages are available to the Conda environment. Example placeholders:

# MaBoSS
# follow MaBoSS/README and upstream docs to install Python bindings and any system dependencies

# NeKo
# follow NeKo/README and upstream docs to install Python dependencies

# PhysiCell (Settings Wrapper)
# follow PhysiCell/README to install necessary Python packages and connection utilities

4. Ensure the MCP configuration points to the correct absolute paths for server.py in each tool folder and to the Python interpreter in the Conda environment. The README in this repo provides example entries that you should adjust to your system paths. Example invocation snippets are shown in the mcp.json example in this repository's guidelines.

5. Restart your editor or reload the MCP configuration so the new servers are registered.

Notes and tips:

• All tools rely on a single Conda environment to simplify dependency management. Keep the environment consistent across maboss, neko, and physicell entries.
• When running, ensure the absolute path to each server.py is correct; relative paths will not work for the MCP client.
• If you encounter path or environment variable issues, verify that PATH includes the Conda environment's bin directory and that CONDA_PREFIX points to the same environment used to launch server.py.
• Depending on your system, you may need to install native dependencies for MaBoSS, NeKo, or PhysiCell (e.g., compilers, system libraries). Consult each upstream project's README for required system prerequisites.
• If you plan to extend with additional servers, follow the same pattern and add a new entry in the mcpServers map with the correct command, args, and any environment needs.