Cbioportal

What Cbioportal does

The Cbioportal MCP server connects AI agents to public cBioPortal cancer genomics data. It supports study discovery, study metadata retrieval, gene identification, and cancer-type lookup. Available studies include public datasets such as TCGA, CPTAC, and MSK, subject to what cBioPortal exposes.

This fits the Biology & Bioinformatics category and is useful for agents that need to identify relevant cancer studies or normalize basic identifiers before a broader analysis workflow. The listed operations return metadata rather than genomic measurements or patient-level analysis results.

How it works

The Cbioportal MCP server is available through two connection models. The hosted option uses the Pipeworx gateway at https://gateway.pipeworx.io/cbioportal/mcp. That endpoint lists the four cBioPortal tools together with shared Pipeworx meta-tools. The gateway can also route questions through ask_pipeworx, which may select tools across the broader Pipeworx catalog.

A local alternative is the @pipeworx/mcp-cbioportal package, launched with npx. It communicates over standard input and output and exposes only this pack’s tools. Both approaches use the same source and tool set for the cBioPortal portion.

For clients or workflows that do not use MCP directly, the gateway also provides HTTP tool calls. For example, POST https://gateway.pipeworx.io/v1/tools/cbioportal_search_studies accepts a JSON body such as {"query":"breast"}.

Setup and configuration

Add the hosted endpoint to an MCP client with a URL configuration:

{
  "mcpServers": {
    "cbioportal": {
      "url": "https://gateway.pipeworx.io/cbioportal/mcp"
    }
  }
}

The README specifically names Claude Desktop, Cursor, and Windsurf as compatible clients. The hosted quick start states that no account is needed for the first calls. To run locally, use:

{
  "mcpServers": {
    "cbioportal": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-cbioportal"]
    }
  }
}

Running npx -y @pipeworx/mcp-cbioportal directly can confirm that the local process starts. No environment variables are documented as required.

Tools and capabilities

• search_studies matches a case-insensitive keyword against study names, study IDs, and cancer-type IDs. Results include IDs, names, cancer types, and sample counts.
• get_study retrieves one study by ID, including its description, cancer type, sample count, PMID, and citation.
• get_gene accepts a HUGO symbol such as TP53 or BRCA1, or an Entrez gene ID, and returns the Entrez ID, symbol, and gene type.
• list_cancer_types returns cBioPortal cancer-type IDs, display names, and parent types. This can help resolve the cancer-type identifiers associated with studies.

Limitations and notes

The Cbioportal MCP server is an independent, unofficial integration and is not affiliated with or endorsed by the upstream provider.

The hosted pack endpoint includes shared gateway tools in addition to the four cBioPortal tools, so its tool count is higher than the standalone package’s count. The exact hosted scope is reported by the MCP initialize response and may change. Use the standalone package when only the listed cBioPortal tools should be visible, or the full Pipeworx gateway when broader catalog access is needed.