Oncology DB

English | 中文

Give your AI assistant the ability to query precision oncology data — gene targets, FDA-approved drugs, clinical trials, and treatment evidence.

An MCP server integrating CIViC, OpenFDA, and ClinicalTrials.gov. Covers 563 gene targets, 9,500+ oncology drugs, and 18,000+ clinical trials.

Installation

Claude Code

claude mcp add --transport http -s user oncology-db https://oncodb.zhiyudao.cn/mcp/mcp

Restart Claude Code. Done.

Claude Desktop / Cursor / Windsurf

Add to your MCP config:

{
  "mcpServers": {
    "oncology-db": {
      "url": "https://oncodb.zhiyudao.cn/mcp/mcp"
    }
  }
}

What Can It Do

Once installed, just ask your AI assistant in natural language:

Query Target Treatments

• "What are the treatment options for KRAS G12C?"
• "Which FDA-approved drugs target EGFR L858R?"
• "Show me resistance evidence for BRAF V600E"

Look Up Drug Details

• "What is the mechanism of action of Sotorasib?"
• "What indications is Pembrolizumab approved for?"
• "How many recruiting clinical trials involve Osimertinib?"

Search Clinical Trials

• "Find recruiting Phase 3 trials for KRAS G12C"
• "Are there any new drug trials for ALK fusions?"

Get Guideline Recommendations

• "What does CIViC assert about BRAF V600E in melanoma?"
• "Show NCCN guideline recommendations for EGFR mutations"

Tools

Tool	Description
search	Search by gene name, drug name, or disease
get_target_treatments	All treatments for a gene-variant target (auto-separates sensitivity vs resistance)
get_drug_detail	Drug details: FDA status, mechanism of action, indications, clinical trials
list_genes	List all gene targets with evidence counts
get_gene_variants	List all variants for a specific gene
search_trials	Search clinical trials by keyword, status, phase
get_stats	Database statistics
get_civic_assertions	CIViC clinical assertions (regulatory approvals, guideline recommendations)

Clinical Accuracy Safeguards

• Sensitivity vs resistance separation — "Supports Resistance" evidence is never shown as a positive treatment option
• Combination therapy context — Clinical-grade (A/B level) combination regimens are labeled (e.g., Cetuximab + Adagrasib)
• Drug type classification — Targeted therapy / Chemotherapy / Immunotherapy / ADC
• Evidence grading — CIViC evidence levels A through E
• Non-drug entity filtering — Excludes drug class names like "BRAF Inhibitor" or "Chemotherapy" from drug results

Data Sources

Source	Content
CIViC	11,232 clinical evidence records + 213 assertions + 1,967 variants
OpenFDA	50,939 FDA drugs + 256,165 drug labels
ClinicalTrials.gov	18,161 oncology clinical trials

Web Interface

A web UI for interactive browsing is also available at https://oncodb.zhiyudao.cn

Disclaimer

This tool is for research reference only and does not constitute medical advice. Clinical decisions should be made in consultation with qualified healthcare professionals.

License

MIT

Feedback

Issues and pull requests: https://github.com/CancerDAO/oncology-db-mcp