playbooks

home / mcp / alphafold mcp server

AlphaFold MCP Server

A comprehensive Model Context Protocol (MCP) server that provides access to the AlphaFold Protein Structure Database through a rich set of tools and resources for protein structure prediction analysis.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "augmented-nature-alphafold-mcp-server": {
      "command": "node",
      "args": [
        "/path/to/alphafold-server/build/index.js"
      ]
    }
  }
}

You can access AlphaFold structure predictions programmatically through this MCP server, enabling retrieval, confidence analysis, batch processing, and visualization-ready exports to streamline your protein research workflows.

How to use

Connect with the AlphaFold MCP server through your MCP client to perform a range of operations. Retrieve structures by UniProt ID with get_structure, check availability with check_availability, and obtain detailed per-residue confidence with get_confidence_scores. Use batch tools like batch_structure_info and batch_download to process multiple proteins at once. Explore comparative features with compare_structures and discover similar proteins via find_similar_structures. Export data for PyMOL or ChimeraX with export_for_pymol and export_for_chimerax, and validate overall quality with validate_structure_quality.

Here are practical usage patterns you can follow: 1) Verify availability for a UniProt ID, 2) Retrieve the structure or metadata, 3) Analyze confidence to identify reliable regions, 4) Prepare visualization-ready exports for your analysis group, 5) Extend to batch workflows for large protein sets, 6) Compare related structures side-by-side to draw structural insights.

How to install

Prerequisites you need before running the server are Node.js and npm. Ensure you have a compatible environment to build and start the MCP server.

# Install dependencies for the MCP server
npm install

# Build the server
npm run build

Configuration and run

Use the following MCP configuration snippet to register the AlphaFold MCP server in your client. This entry uses a local stdio setup that runs the server from a built index file.

{
  "mcpServers": {
    "alphafold-server": {
      "command": "node",
      "args": ["/path/to/alphafold-server/build/index.js"]
    }
  }
}

Starting the server directly

If you prefer to start the server manually after building, you can run the start command directly from the build output.

# Start the server directly
node build/index.js

Notes on usage and capabilities

This MCP server provides a comprehensive set of tools to work with AlphaFold predictions, including per-residue confidence scores, region analysis, batch processing, comparative structure analysis, and export capabilities for PyMOL and ChimeraX. All data is retrieved from the AlphaFold API to ensure you have up-to-date predictions and metadata for your research.

Tools available in this MCP server

The server exposes a broad set of tools that you can call via your MCP client. These tools perform retrieval, analysis, batch processing, comparison, and export tasks.

Security and environment

Run the server in a trusted environment. The standard setup does not require extra environment variables beyond what you configure for your runtime. Monitor API usage and respect rate limits when performing batch operations.

Available tools

get_structure

Retrieve AlphaFold structure prediction for a specific UniProt ID. Supports output formats pdb, cif, bcif, or json.

download_structure

Download AlphaFold structure file in the specified format (pdb, cif, or bcif).

check_availability

Check if a structure prediction is available for a given UniProt ID.

search_structures

Search available AlphaFold structures by protein name or gene, with optional organism filtering and result size.

list_by_organism

List all available structures for a specific organism with an optional result size.

get_organism_stats

Get statistics about AlphaFold coverage for an organism.

get_confidence_scores

Retrieve per-residue confidence scores for a structure prediction with optional thresholding.

analyze_confidence_regions

Analyze confidence score distribution and identify high/low confidence regions.

get_prediction_metadata

Retrieve metadata about a prediction including version, date, and quality metrics.

batch_structure_info

Get structure information for multiple proteins in a single call, with output format json or summary.

batch_download

Download multiple structure files in a single operation.

batch_confidence_analysis

Analyze confidence scores for multiple proteins in one batch.

compare_structures

Compare multiple AlphaFold structures side-by-side for analysis.

find_similar_structures

Find AlphaFold structures similar to a given protein, with optional organism filter.

get_coverage_info

Get information about sequence coverage in the AlphaFold prediction.

validate_structure_quality

Assess the overall quality and reliability of a prediction.

export_for_pymol

Export structure data formatted for PyMOL visualization, with optional confidence coloring.

export_for_chimerax

Export structure data formatted for ChimeraX visualization, with optional confidence coloring.

get_api_status

Check AlphaFold API status and database statistics.