Unofficial PubChem MCP Server

What is Unofficial PubChem MCP Server?

The Unofficial PubChem MCP Server is a comprehensive Model Context Protocol (MCP) server for accessing the PubChem chemical database. It provides access to over 110 million chemical compounds with extensive molecular properties, bioassay data, and chemical informatics tools. The server is designed for researchers, chemists, and drug discovery professionals.

How to use Unofficial PubChem MCP Server?

Install dependencies, build, and run the server using npm commands: npm install, npm run build, npm start. Invoke tools via JSON-RPC requests to the MCP server with tool names and arguments (e.g., search_compounds, get_compound_info). Batch processing is available via the batch_compound_lookup tool, handling up to 200 compounds.

Key features of Unofficial PubChem MCP Server

• Search compounds by name, CAS, formula, or identifier using 6 search tools.
• Perform Tanimoto similarity, substructure, and superstructure searches.
• Calculate molecular descriptors, ADMET predictions, and drug-likeness.
• Access bioassay data, activity profiles, and target-based searches.
• Retrieve safety data, toxicity info, and regulatory classifications.
• Batch process up to 200 compounds for property lookups.

Use cases of Unofficial PubChem MCP Server

• Drug discovery workflow: search compounds, assess drug-likeness, check safety profiles.
• Chemical similarity analysis: find similar compounds and compare properties.
• Target discovery: find small molecule ligands for protein targets.
• Safety and regulatory assessment: retrieve GHS hazards and toxicity data.
• Cross-database integration: combine with UniProt, ChEMBL, and STRING servers.

FAQ from Unofficial PubChem MCP Server

What are the API rate limits?

PubChem API allows 5 requests per second and a maximum of 400 requests per minute. No API key is required. Respectful usage is encouraged.

What data sources does the server access?

The server queries PubChem, which contains 110+ million compounds, 1.5+ million bioassays, computed chemical properties, 3D conformers, safety classifications, and cross-references to 500+ external databases.

How do I integrate this server with other MCP servers?

The server works with UniProt (find protein targets → ligands), ChEMBL (analyze bioactivity), and STRING (protein interactions → chemical modulators) to create a complete chemical informatics pipeline.

What is the maximum number of compounds for batch processing?

The batch_compound_lookup tool supports up to 200 compounds per call. Invalid compound IDs result in clear error messages.

Does the server require authentication?

No. The PubChem API does not require an API key. The server includes built-in rate limiting compliance and TypeScript validation for all inputs.