MCP Server#

The BDP MCP (Model Context Protocol) server exposes the biological knowledge graph to AI agents — Claude, GPT, and any MCP-compatible client. Agents can traverse gene → disease → phenotype → pathway → drug → literature relationships using structured tools.

What Is MCP?#

Model Context Protocol is an open standard for connecting AI models to external tools and data sources. BDP implements an MCP server so that AI assistants can query the knowledge graph directly during a conversation.

Transport Modes#

BDP MCP supports two transports:

| Mode | Use case | |---|---| | stdio | Local use with Claude Desktop, Cursor, or any MCP client | | HTTP (SSE) | Remote or cloud-hosted deployments |

Connecting via stdio#

Add to your MCP client configuration (e.g. Claude Desktop config.json):

json

{
  "mcpServers": {
    "bdp": {
      "command": "bdp-mcp",
      "args": ["--server-url", "https://bdp.dev"]
    }
  }
}

Connecting via HTTP#

Point your MCP client at the SSE endpoint:

Code

https://bdp.dev/mcp/sse

What Agents Can Do#

Once connected, an AI agent can:

Look up gene information, coordinates, and cross-references
Find diseases associated with a gene
Explore phenotypes linked to a disease
Identify drugs and their indications or adverse events
Traverse pathways containing a protein
Search literature by entity or semantic query
Retrieve protein structures (PDB)
Run graph traversal queries

See the Tools Reference for the full list of available tools.

Example Agent Interaction#

User: What drugs target EGFR and what are their approved indications?

The agent calls:

get_gene — resolve EGFR to its identifiers
get_gene_drugs — find drugs targeting EGFR
get_drug_indications (per drug) — retrieve approved uses

All without leaving the conversation.

Documentation