Alphafold MCP Server

🚀 AlphaFold MCP Server

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

🚀 Quick Start

This MCP server enables seamless integration with AlphaFold's vast collection of protein structure predictions, offering tools for structure retrieval, confidence analysis, batch processing, and visualization preparation. It's perfect for researchers, bioinformaticians, and structural biologists working with predicted protein structures.

✨ Features

🧬 Core Structure Tools

• Structure Retrieval: Retrieve AlphaFold predictions by UniProt ID.
• Multi-format Downloads: Support for PDB, CIF, BCIF, and JSON formats.
• Availability Checking: Verify if predictions exist for specific proteins.

🔍 Search & Discovery

• Structure Search: Find proteins by name, gene, or organism.
• Organism Browsing: List all available structures for specific species.
• Coverage Statistics: Get comprehensive organism - level statistics.

📊 Confidence & Quality Analysis

• Per - residue Confidence: Obtain detailed confidence scores for each amino acid.
• Region Analysis: Identify high/low confidence structural regions.
• Quality Validation: Assess overall prediction reliability.

⚡ Batch Processing

• Bulk Retrieval: Process multiple proteins simultaneously.
• Batch Downloads: Efficiently download multiple structures.
• Parallel Analysis: Conduct confidence analysis for protein sets.

🔬 Comparative Analysis

• Structure Comparison: Perform side - by - side analysis of multiple proteins.
• Similarity Search: Find structurally related proteins.
• Coverage Comparison: Analyze prediction completeness.

🎨 Visualization Integration

• PyMOL Scripts: Use ready - to - use visualization scripts.
• ChimeraX Integration: View confidence - colored structures.
• Custom Export Formats: Enjoy flexible data export options.

📦 Installation

# Clone or create the server directory
npm install

# Build the server
npm run build

💻 Usage Examples

As MCP Server

Add to your MCP configuration:

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

Direct Usage

# Start the server
npm start

# Or run directly
node build/index.js

📚 Documentation

Available Tools

Core Structure Tools

get_structure

Retrieve AlphaFold structure prediction for a specific UniProt ID.

Parameters:

• uniprotId (required): UniProt accession (e.g., "P21359", "Q8N726")
• format (optional): Output format - "pdb", "cif", "bcif", or "json" (default: "json")

Example:

{
  "uniprotId": "P04637",
  "format": "json"
}

download_structure

Download AlphaFold structure file in specified format.

Parameters:

• uniprotId (required): UniProt accession
• format (optional): File format - "pdb", "cif", or "bcif" (default: "pdb")

check_availability

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

Parameters:

• uniprotId (required): UniProt accession to check

Search & Discovery Tools

search_structures

Search for available AlphaFold structures by protein name or gene.

Parameters:

• query (required): Search term (protein name, gene name, etc.)
• organism (optional): Filter by organism
• size (optional): Number of results (1 - 100, default: 25)

list_by_organism

List all available structures for a specific organism.

Parameters:

• organism (required): Organism name (e.g., "Homo sapiens", "Escherichia coli")
• size (optional): Number of results (1 - 100, default: 50)

get_organism_stats

Get statistics about AlphaFold coverage for an organism.

Parameters:

• organism (required): Organism name

Confidence & Quality Tools

get_confidence_scores

Get per - residue confidence scores for a structure prediction.

Parameters:

• uniprotId (required): UniProt accession
• threshold (optional): Confidence threshold (0 - 100)

analyze_confidence_regions

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

Parameters:

• uniprotId (required): UniProt accession

get_prediction_metadata

Get metadata about the prediction including version, date, and quality metrics.

Parameters:

• uniprotId (required): UniProt accession

Batch Processing Tools

batch_structure_info

Get structure information for multiple proteins simultaneously.

Parameters:

• uniprotIds (required): Array of UniProt accessions (max 50)
• format (optional): Output format - "json" or "summary" (default: "json")

batch_download

Download multiple structure files.

Parameters:

• uniprotIds (required): Array of UniProt accessions (max 20)
• format (optional): File format - "pdb" or "cif" (default: "pdb")

batch_confidence_analysis

Analyze confidence scores for multiple proteins.

Parameters:

• uniprotIds (required): Array of UniProt accessions (max 30)

Comparative Analysis Tools

compare_structures

Compare multiple AlphaFold structures for analysis.

Parameters:

• uniprotIds (required): Array of UniProt accessions to compare (2 - 10)

find_similar_structures

Find AlphaFold structures similar to a given protein.

Parameters:

• uniprotId (required): Reference UniProt accession
• organism (optional): Filter by organism

Coverage & Completeness Tools

get_coverage_info

Get information about sequence coverage in the AlphaFold prediction.

Parameters:

• uniprotId (required): UniProt accession

validate_structure_quality

Validate and assess the overall quality of an AlphaFold prediction.

Parameters:

• uniprotId (required): UniProt accession

Export & Integration Tools

export_for_pymol

Export structure data formatted for PyMOL visualization.

Parameters:

• uniprotId (required): UniProt accession
• includeConfidence (optional): Include confidence score coloring (default: true)

export_for_chimerax

Export structure data formatted for ChimeraX visualization.

Parameters:

• uniprotId (required): UniProt accession
• includeConfidence (optional): Include confidence score coloring (default: true)

get_api_status

Check AlphaFold API status and database statistics.

Parameters: None

Available Resources

Resource Templates

alphafold://structure/{uniprotId}

MIME Type: application/json Description: Complete AlphaFold structure prediction for a UniProt ID

alphafold://pdb/{uniprotId}

MIME Type: chemical/x-pdb Description: PDB format structure file for a UniProt ID

alphafold://confidence/{uniprotId}

MIME Type: application/json Description: Per - residue confidence scores for a structure prediction

alphafold://summary/{organism}

MIME Type: application/json Description: Summary of all available structures for an organism

Example Workflows

Basic Structure Analysis

// 1. Check if structure is available
await use_mcp_tool("alphafold-server", "check_availability", {
  uniprotId: "P04637",
});

// 2. Get structure metadata
await use_mcp_tool("alphafold-server", "get_prediction_metadata", {
  uniprotId: "P04637",
});

// 3. Analyze confidence scores
await use_mcp_tool("alphafold-server", "get_confidence_scores", {
  uniprotId: "P04637",
  threshold: 70,
});

Comparative Study

// Compare multiple related proteins
await use_mcp_tool("alphafold-server", "compare_structures", {
  uniprotIds: ["P04637", "P53350", "P63151"],
});

// Batch confidence analysis
await use_mcp_tool("alphafold-server", "batch_confidence_analysis", {
  uniprotIds: ["P04637", "P53350", "P63151"],
});

Visualization Preparation

// Export for PyMOL with confidence coloring
await use_mcp_tool("alphafold-server", "export_for_pymol", {
  uniprotId: "P04637",
  includeConfidence: true,
});

// Export for ChimeraX
await use_mcp_tool("alphafold-server", "export_for_chimerax", {
  uniprotId: "P04637",
  includeConfidence: true,
});

Organism - wide Analysis

// Get human protein statistics
await use_mcp_tool("alphafold-server", "get_organism_stats", {
  organism: "Homo sapiens",
});

// List available structures
await use_mcp_tool("alphafold-server", "list_by_organism", {
  organism: "Homo sapiens",
  size: 100,
});

API Reference

The server connects to the AlphaFold API at https://alphafold.ebi.ac.uk/api/ and provides structured access to:

• Structure Predictions: Complete protein structure data
• Confidence Scores: Per - residue reliability metrics
• Metadata: Prediction versions, dates, and quality information
• Cross - references: Links to other databases and resources

🔧 Technical Details

Error Handling

The server includes comprehensive error handling for:

• Invalid UniProt IDs
• Missing structure predictions
• API connectivity issues
• Rate limiting and timeouts
• Malformed requests

Rate Limiting

Please be mindful of API usage:

• Batch operations are limited to reasonable sizes
• Large requests are automatically chunked
• Built - in delays prevent API overload

🤝 Contributing

Contributions are welcome! Please ensure:

• TypeScript type safety
• Comprehensive error handling
• Documentation for new features
• Testing with real AlphaFold data

👋 About

Developed by Augmented Nature - augmentednature.ai

Augmented Nature specializes in creating AI - powered tools and infrastructure for scientific research and data analysis.

📄 License

MIT License - see LICENSE file for details.

📖 Citation

If you use this MCP server in your research, please cite:

• AlphaFold Database: https://alphafold.ebi.ac.uk/
• Model Context Protocol: https://modelcontextprotocol.io/

If you use this project in your research or publications, please cite it as follows:

@misc{alphafoldmcp2025,
author = {Moudather Chelbi},
title = {AlphaFold MCP Server},
year = {2025},
howpublished = {https://github.com/Augmented-Nature/AlphaFold-MCP-Server},
note = {Accessed: 2025-06-29}
}

⚠️ Important Note

This server provides a convenient interface to AlphaFold data but does not store or cache structure data. All data is retrieved directly from the official AlphaFold API.

💡 Usage Tip

For issues and questions:

• Check the AlphaFold API documentation
• Review error messages for debugging hints
• Ensure UniProt IDs are valid and current