rosetta

Rosetta MCP Server exposes a set of tools to work with Rosetta and PyRosetta through a Cursor-compatible MCP interface. It lets clients run RosettaScripts, validate and translate Rosetta XML, and browse common movers, filters, and selectors. Optional PyRosetta tools provide scoring and introspection capabilities, while environment helpers assist with local Python environments and installation. You can use commands like run_rosetta_scripts to execute RosettaScripts against your input structures, rosetta_scripts_schema to generate the RosettaScripts schema, and validate_xml to check XML syntax. Additionally, tools such as pyrosetta_score and pyrosetta_introspect enable scoring a PDB with PyRosetta and exploring PyRosetta classes and signatures.

Prerequisites:

• Node.js and npm installed on your system
• Internet access to fetch the MCP server wrapper from npm

Installation steps (recommended):

1. Install the MCP server wrapper globally via npm

npm install -g rosetta-mcp-server

2. Run the server using npx (as configured in mcp_config)

# This will start the rosetta MCP server using the wrapper package
# The exact command may vary depending on your environment
rosetta-mcp-server --help

3. First-time PyRosetta setup occurs automatically when you invoke PyRosetta tools. This can take 10–30 minutes and will display progress indicators. No manual PyRosetta installation is required upfront.

Optional: Install from source

1. Ensure prerequisites are installed (macOS example):

brew install node
python3 --version

2. Clone and install globally:

git clone https://github.com/yourusername/rosetta-mcp-server.git
cd rosetta-mcp-server
npm install -g .

4. Verify installation by running the server or listing available tools (as per your environment).

Notes and tips:

• PyRosetta installation is automatic on first use of PyRosetta tools (pyrosetta_score, pyrosetta_introspect). Expect 10–30 minutes for the initial download/compilation.
• The server includes smart environment handling to resolve conflicts (e.g., Conda/anaconda). It can create a clean Python environment if needed and store it under a dedicated path.
• You can configure the server in Cursor by editing ~/.cursor/mcp.json and adding the rosetta entry with ROSETTA_BIN and PYTHON_BIN if you have a specific RosettaScripts path or Python interpreter.
• For RosettaTools usage, you can run run_rosetta_scripts with inputs like xml_path, input_pdb, and out_dir. You can override the executable with exe_path or rely on ROSETTA_BIN or PATH discovery.
• The XML to PyRosetta translation feature can help convert RosettaScripts XML into equivalent PyRosetta Python code for easier experimentation and scripting.