ProteinAtlas

The ProteinAtlas MCP Server provides programmatic access to Human Protein Atlas data through a consolidated MCP interface. It offers core data modalities including basic protein information, tissue and brain expression, blood expression, subcellular localization, pathology/cancer data, and antibody information, along with advanced search capabilities and batch processing for handling multiple queries at once. You can query for proteins by gene symbol or description, retrieve tissue- and brain-specific expression data, explore subcellular localization patterns, and fetch pathology-related markers and antibody details. The server exposes a collection of tools such as search_proteins, get_protein_info, get_tissue_expression, get_subcellular_location, get_pathology_data, and batch_protein_lookup, enabling both simple lookups and complex, multi-criteria searches via the MCP protocol.

To use the server from an MCP client, configure the MCP client to point to the proteina atlas MCP server instance. The client can then invoke the available tools with the specified parameters (e.g., query strings, formats, and filters) and receive structured results in formats like json or tsv. This setup enables streamlined integration into research pipelines, dashboards, or automated annotation workflows that rely on Human Protein Atlas data for expression, localization, and pathology context.

Prerequisites:

• Node.js 18 or higher
• npm or yarn package manager

Setup steps:

1. Clone or download the server repository
2. Install dependencies

cd proteinatlas-server
npm install

3. Build the server

npm run build

4. Run the server locally (example)

npm start
# or
node build/index.js

The server should now be ready to accept MCP requests from configured clients.

Tips and notes:

• Ensure you are using Node.js 18+ to satisfy the server's runtime requirements.
• When integrating with an MCP client, point the client at the server's executable path as shown in the integration example, adjusting the path to your actual build location.
• If you enable batch_protein_lookup or advanced_search, consider setting maxResults to an upper bound that your deployment can handle to avoid heavy load.
• Prepare format-specific expectations (json, tsv, etc.) in your client to avoid parsing surprises.
• Monitor for large query responses; implement pagination or result chunking if supported by the client to maintain responsiveness.