rosetta_kic_mcp

Rosetta KIC MCP provides tools for cyclic peptide analysis using Rosetta's Kinematic Closure (KIC) and GeneralizedKIC protocols. Through the MCP server interface you can discover available tools, submit long-running jobs (such as cyclic peptide closure, structure prediction, and loop modeling), and track their progress. Core capabilities include closing linear peptides into cyclic structures, predicting 3D structures from sequences, modeling flexible loops, and batch processing for high-throughput workflows. The server exposes both quick, synchronous operations (validating sequences/structures and getting server info) and asynchronous, long-running tasks that return a job_id for monitoring.

To use it, connect via Claude Code or Gemini CLI with the configured mcpServer entry named rosetta-kic-mcp. Tools are organized under quick operations (sync) and long-running tasks (submit). Typical workflows involve validating inputs, submitting structure prediction or closure jobs, and then checking job status or retrieving results when complete. The available endpoints include validate_peptide_structure, validate_peptide_sequence, get_server_info, submit_cyclic_peptide_closure, and structure prediction/loop modeling utilities, among others documented in the Available Tools section of the MCP server README.

Prerequisites:

• A Unix-like environment with Python 3.8+ (recommended 3.12 as per README example)
• Conda or Mamba for environment management (optional but recommended)
• Basic Git and shell access

Installation steps:

1. Clone or download the repository containing the MCP server:

   • git clone https://github.com/<your-org>/ CycPepMCP
   • cd CycPepMCP/tool-mcps/rosetta_kic_mcp

2. Set up a dedicated Python environment (example using conda/mamba):

   • mamba create -p ./env python=3.12 pip numpy pandas -y
   • conda activate ./env (or: mamba activate ./env)

3. Install required Python packages and MCP framework:

   • mamba install -p ./env -c conda-forge rdkit -y
   • mamba install -p ./env -c conda-forge biopython -y
   • mamba run -p ./env pip install loguru click tqdm
   • mamba run -p ./env pip install fastmcp
   • pip install pyrosetta --find-links https://west.rosettacommons.org/pyrosetta/quarterly/release

4. Verify installation and start the MCP server as described in the README, using the Python server entry point:

   • cd /home/xux/Desktop/CycPepMCP/CycPepMCP/tool-mcps/rosetta_kic_mcp
   • mamba activate ./env (if not already active)
   • python src/server.py (or use your preferred invocation)

5. Optional Claude Code or Gemini CLI integration:

   • Follow the instructions in the README to register the server under the mcpServers setting for Claude Code or Gemini CLI.

Tips and common issues:

• Ensure the Python environment used to run the MCP server matches the one used to install dependencies (path in the mcp config should point to the same env).
• If using Claude Code or Gemini CLI, keep the absolute path to the server script consistent with your system layout.
• For long-running tasks, monitor job_status via the returned job_id and adjust runtime/limits in your tool parameters as needed.
• If RDKit or PyRosetta fail to load, verify that library binaries are compatible with your OS and Python version; conda-forge channels are typically the safest route.
• You can customize environment variables in the MCP setup if your deployment requires specific paths or data directories (e.g., data roots, cache dirs).
• The server supports both quick validation calls and longer computational jobs; use quick operations for input checks before submitting heavier tasks.