BioMCP: Biomedical Model Context Protocol

Name: BioMCP: Biomedical Model Context Protocol
Author: genomoncology

@genomoncology

リポジトリホームページ

BioMCP: Biomedical Model Context Protocol について

#bioinformatics #ai #genomics #mcp #pubmed #pubmed-central #medical #clinical-trials #llm #model-context-protocol #mcp-server

基本情報

カテゴリ

その他

ライセンス

MIT

ランタイム

rust

トランスポート

stdio

公開者

genomoncology

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "biomcp": {
      "command": "uv",
      "args": [
        "tool",
        "install",
        "biomcp-cli"
      ]
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is BioMCP?

BioMCP is a command-line and MCP server tool that unifies search and retrieval across dozens of biomedical sources (PubTator3, Europe PMC, ClinVar, UniProt, etc.) behind a single grammar. It is designed for researchers, clinicians, and AI agents who need compact, evidence-oriented results without rebuilding workflows for each source.

How to use BioMCP?

Install the biomcp binary via the install script or uv tool install biomcp-cli, then run biomcp serve to start the MCP stdio server. Use commands such as search, get, suggest, enrich, batch, or study directly in the terminal, or register the server with Claude Desktop, Claude Code, or Codex. Optional API keys improve rate limits and unlock additional data.

Key features of BioMCP

Search literature across PubTator3, Europe PMC, and Semantic Scholar
Pivot between genes, variants, drugs, diseases, and more
Route questions to shipped worked examples with suggest
Analyze local study datasets with native terminal charts
Follow paper trails with citations, references, and recommendations
Enrich gene sets via g:Profiler and batch multiple focused queries

Use cases of BioMCP

Literature discovery: search articles by gene and disease filters
Variant interpretation: get ClinVar, gnomAD, and OncoKB data for a variant
Drug interaction scanning: query drugs and their adverse events or approvals
Gene-set enrichment: run enrichment analysis on a list of genes
Cross-entity pivoting: move from a disease to its known drugs, trials, or articles

FAQ from BioMCP

What biomedical sources does BioMCP query?

BioMCP queries over 40 upstream providers including PubTator3, Europe PMC, MyGene.info, UniProt, ClinVar, ClinicalTrials.gov, OpenFDA, MyDisease.info, Reactome, and many more, depending on the entity and command.

Do I need API keys to use BioMCP?

Most commands work without credentials. Optional keys such as NCBI_API_KEY, S2_API_KEY, OPENFDA_API_KEY, and ONCOKB_TOKEN improve rate limits or unlock additional data sections (e.g., Semantic Scholar TLDRs, OncoKB variant evidence).

How do I pivot between different entities?

Use built-in cross-entity helpers like biomcp variant trials "BRAF V600E" or biomcp disease drugs melanoma. These commands pivot from one entity to related entities without rebuilding filters.

What transport protocols does BioMCP support?

BioMCP supports stdio (local MCP server) and HTTP (remote server via biomcp serve-http). The HTTP server exposes health probes and a Streamable HTTP MCP endpoint.

Are there any known limitations in data coverage?

References and recommendations for articles can be empty for paywalled papers because of publisher elision in Semantic Scholar upstream coverage. Additionally, some detail sections (e.g., diagnostics, funding) remain opt‑in and are not included in the all shortcut.