Unofficial DB MCP Server

Model Context Protocol (MCP) server providing access to the comprehensive DB pharmaceutical database (17,430+ drugs).

Features

• Single unified tool with 16 methods

• High-performance SQLite backend: <10ms queries, ~50-100MB memory usage

• Access to 17,430 drug records (13,166 small molecules + 4,264 biotech)

• Comprehensive pharmaceutical data including:

  • Drug names, descriptions, classifications

  • Clinical indications and mechanisms of action

  • Chemical structures (SMILES, InChI)

  • Drug interactions and contraindications

  • Target proteins, enzymes, carriers, transporters

  • Metabolic pathways

  • Market products and regulatory information

  • Pharmacokinetics (half-life search) and toxicity data

  • Salt forms and external database identifiers

  • Drug similarity search

Related MCP server: Healthcare MCP Server

Installation

# Clone and install dependencies
cd drugbank-mcp-server
npm install

Obtain the DB database:

1. Download the full database ZIP from your DB account

2. Extract the XML and place it at data/full database.xml

# Build the SQLite database and copy source files
npm run build

Usage

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

Tool

Single unified tool with multiple methods accessed via the method parameter.

Methods

1. search_by_name

Search drugs by name (supports partial matching).

Note: DB uses chemical names as primary identifiers. Search for "Acetylsalicylic" to find aspirin, "Ibuprofen" not "Advil", etc.

{
  "method": "search_by_name",
  "query": "Acetylsalicylic",
  "limit": 20
}

2. get_drug_details

Get complete drug information by DB ID.

{
  "method": "get_drug_details",
  "drugbank_id": "EXPT00475"
}

Returns full drug record including:

• All identifiers (DB ID, CAS, UNII)

• Clinical information (indication, mechanism, toxicity)

• Pharmacokinetics (absorption, metabolism, half-life)

• Chemical properties

• Interactions (drug-drug, food)

• Targets and enzymes

3. search_by_indication

Find drugs by medical indication.

{
  "method": "search_by_indication",
  "query": "pain",
  "limit": 20
}

4. search_by_target

Find drugs by target protein/enzyme.

{
  "method": "search_by_target",
  "target": "COX-2",
  "limit": 20
}

5. get_drug_interactions

Get all drug-drug interactions for a specific drug.

{
  "method": "get_drug_interactions",
  "drugbank_id": "DB00945"
}

6. search_by_atc_code

Search by ATC (Anatomical Therapeutic Chemical) classification code.

{
  "method": "search_by_atc_code",
  "code": "N02BA",
  "limit": 20
}

7. get_pathways

Get metabolic pathways for a drug.

{
  "method": "get_pathways",
  "drugbank_id": "DB00945"
}

8. search_by_structure

Search by chemical structure (SMILES or InChI).

{
  "method": "search_by_structure",
  "smiles": "CC(=O)Oc1ccccc1C(=O)O",
  "limit": 20
}

9. get_products

Get market products for a drug (brand names, manufacturers).

{
  "method": "get_products",
  "drugbank_id": "DB00945",
  "country": "US"
}

10. search_by_category

Search drugs by therapeutic category.

{
  "method": "search_by_category",
  "category": "Anti-inflammatory",
  "limit": 20
}

11. get_external_identifiers

Get cross-database identifiers (PubChem, ChEMBL, KEGG, RxCUI, etc.) and structure identifiers.

{
  "method": "get_external_identifiers",
  "drugbank_id": "DB02351"
}

12. search_by_halflife

Find drugs by elimination half-life range (in hours). Useful for dosing considerations.

{
  "method": "search_by_halflife",
  "min_hours": 12,
  "max_hours": 48,
  "limit": 20
}

13. get_similar_drugs

Find drugs similar to a reference drug based on shared targets, categories, and ATC codes. Uses Jaccard similarity scoring.

{
  "method": "get_similar_drugs",
  "drugbank_id": "APRD00003",
  "limit": 20
}

Returns similarity scores with breakdown by:

• target_similarity: Shared protein/enzyme targets (50% weight)

• category_similarity: Shared therapeutic categories (30% weight)

• atc_similarity: Shared ATC classification codes (20% weight)

14. search_by_carrier

Find drugs by carrier protein (proteins that transport drugs in the body, like albumin).

{
  "method": "search_by_carrier",
  "carrier": "Albumin",
  "limit": 20
}

15. search_by_transporter

Find drugs by transporter protein (membrane proteins that move drugs across cell membranes).

{
  "method": "search_by_transporter",
  "transporter": "P-glycoprotein",
  "limit": 20
}

16. get_salts

Get salt forms for a drug (different chemical forms like hydrochloride, sulfate).

{
  "method": "get_salts",
  "drugbank_id": "DB00007"
}

Example Queries with Claude

Once configured, you can ask Claude:

• "Find information about acetylsalicylic acid using DB" (aspirin's chemical name)

• "Search DB for drugs containing ibuprofen"

• "What drugs interact with warfarin?"

• "Show me all drugs that target COX-2"

• "Find drugs used for treating hypertension"

• "What are the metabolic pathways for acetylsalicylic acid?"

• "Find drugs similar to Nelfinavir" (HIV protease inhibitor)

• "What are the external identifiers for Bivalirudin?"

• "Find drugs with a half-life between 12 and 24 hours"

• "What drugs are carried by albumin?"

• "Find drugs transported by P-glycoprotein"

• "What salt forms are available for leuprolide?"

Note: Use chemical/generic names (acetylsalicylic acid, ibuprofen, acetaminophen) rather than brand names (Aspirin, Advil, Tylenol) for best results.

Performance

SQLite Mode (Default)

• All queries: <10ms

• Memory usage: ~50-100MB

• Database size: 31.1MB (98% reduction from 1.5GB XML)

• FTS5 full-text search: Fast name/indication lookups

XML Fallback Mode

• First query: ~30-60 seconds (loads entire 1.5GB XML into memory)

• Subsequent queries: <500ms (cached in memory)

• Memory usage: ~2-3GB when database is loaded

Data Source

• Database: DB (Full Database)

• Records: 17,430 drugs (13,166 small molecules + 4,264 biotech)

License

MIT