BioMCP

Content

BioMCP is a biomedical Model Context Protocol server and command-line tool for evidence retrieval, entity lookup, and local study analysis. It lets Claude use the biomcp command grammar through MCP to search articles, genes, variants, clinical trials, diagnostics, drugs, diseases, pathways, proteins, adverse events, pharmacogenomics, GWAS records, phenotypes, and approved local study datasets.

The project focuses on read-only biomedical discovery. It aggregates public and optional authenticated data sources such as PubMed, PubTator3, Europe PMC, ClinicalTrials.gov, NCI CTS, MyGene.info, MyVariant.info, ClinVar, MyChem.info, OpenFDA, Reactome, UniProt, CIViC, OncoKB, g:Profiler, and Semantic Scholar.

Source Review

• https://github.com/genomoncology/biomcp
• https://github.com/genomoncology/biomcp/blob/main/README.md
• https://pypi.org/pypi/biomcp-cli/json
• https://github.com/genomoncology/biomcp/blob/main/pyproject.toml
• https://github.com/genomoncology/biomcp/blob/main/Cargo.toml
• https://github.com/genomoncology/biomcp/blob/main/manifest.json
• https://github.com/genomoncology/biomcp/blob/main/LICENSE

These sources were reviewed on 2026-06-06. Prefer the live repository, README, PyPI metadata, Python package metadata, Rust package metadata, MCP manifest, and license file for current install commands, supported entities, tool behavior, optional credentials, and licensing.

Features

• Python package biomcp-cli with a Rust-backed biomcp binary.
• Stdio MCP server launched with biomcp serve.
• Read-only biomedical command grammar exposed through a biomcp MCP tool.
• Literature search across PubTator3 and Europe PMC, with article detail, citation, reference, recommendation, and entity workflows.
• Gene, variant, disease, drug, diagnostic, pathway, protein, adverse-event, pharmacogenomics, GWAS, and phenotype lookup workflows.
• Clinical trial search through ClinicalTrials.gov API v2 and NCI CTS sources.
• Local study analytics for approved downloaded datasets, including cohort, survival, comparison, co-occurrence, and chart workflows.
• Optional API keys for higher-rate or gated integrations.
• MIT license.

Installation

Install the package, then configure the MCP server:

uv tool install biomcp-cli

{
  "mcpServers": {
    "biomcp": {
      "command": "biomcp",
      "args": ["serve"]
    }
  }
}

After restarting the MCP client, ask Claude to search or summarize an approved biomedical question and cite the returned source records.

Use Cases

• Search biomedical literature for a gene, variant, disease, drug, or phenotype.
• Compare article evidence with clinical trial records for a research question.
• Retrieve structured detail pages for genes, variants, drugs, diseases, or trials.
• Explore known gene, variant, disease, pathway, and drug relationships.
• Summarize public clinical trial status and eligibility criteria for review.
• Run approved local study analytics on de-identified datasets.
• Generate evidence-backed biomedical research notes with source links.

Safety and Privacy

BioMCP should be treated as a research aid. Its outputs are not medical advice and should not be used directly for diagnosis, treatment, eligibility determinations, regulatory submissions, or safety decisions without qualified human review and source verification.

Biomedical searches can expose sensitive research direction, disease interests, drug names, variants, trial queries, and cohort context to third-party APIs. Local study workflows can also handle sensitive derived data and generated charts. Use only approved datasets, protect optional API keys, and avoid PHI or patient-identifiable records unless your environment and approvals explicitly permit that use.

Duplicate Check

Existing content includes paper-search and PubMed-oriented MCP entries, plus life-sciences agent content. This entry is distinct because it covers genomoncology/biomcp, a unified biomedical MCP server and CLI spanning literature, genes, variants, trials, diagnostics, drugs, diseases, pathways, proteins, adverse events, pharmacogenomics, GWAS, phenotypes, and local study analytics. No matching BioMCP source URL or dedicated BioMCP entry was found in content/mcp.