Awesome MCP › Other MCP Servers

Nexgene-Research/nexonco-mcp

⭐ 64 Python repository created 2025-04-13

Nexonco is a Model Context Protocol server, published on PyPI as nexonco-mcp, that gives an LLM client access to clinical evidence from CIViC, the Clinical Interpretation of Variants in Cancer database. It is built by Nexgene Research, licensed MIT, and registered with the NANDA server directory. The server exposes a single tool, search_clinical_evidence, which accepts a set of optional filters and returns a formatted report rather than raw records. The filters are disease name, therapy or drug name, molecular profile covering gene or variant, phenotype name, evidence type such as predictive or diagnostic, evidence direction such as supports, and a boolean that restricts results to strong evidence, defined as a rating above three on a five point scale. Because every parameter is optional, a query can be as broad or as narrow as the question requires. The returned report has four fixed sections. Summary statistics give the total number of evidence items, the average evidence rating, and the three most frequent diseases, genes, variants, therapies and phenotypes with their counts. The next section lists the ten highest rated evidence entries with the disease, phenotype, gene and variant, therapy, description, type, direction and rating for each. The third section gives citations and source URLs for those ten entries, so the model can attribute what it reports. The fourth is a disclaimer stating that the tool is for research purposes only and is not a substitute for professional medical advice, diagnosis or treatment, a point the README repeats. Installation uses either uv or Docker, and setup guides in the docs directory cover Claude Desktop integration and local registration with a NANDA host. Sample questions in the README show the intended use: finding predictive evidence for a therapy and mutation combination, or checking which therapies are linked to a given cancer.

https://github.com/Nexgene-Research/nexonco-mcp

mcpbioinformaticsoncologycivicclinical-evidencepythonresearch