Enrichr MCP Server

Overview

The Enrichr MCP server turns the rich, publicly available gene‑set enrichment resource from Maayan Lab into a lightweight, AI‑friendly service. By exposing Enrichr’s API through the Model Context Protocol, it allows language models to request statistically significant enrichment results for a user’s gene list without any manual web scraping or data wrangling. This solves the problem of bridging static biological databases with dynamic AI workflows: developers can now embed sophisticated pathway, disease, and drug‑target analyses directly into conversational agents or code assistants.

When a client sends a gene list, the server queries all configured Enrichr libraries and filters the results to only include entries with a corrected p‑value below 0.05. The response is concise and machine‑readable, enabling the LLM to interpret, summarize, or act on the findings. Because the server handles library selection and statistical filtering internally, developers can focus on higher‑level logic—such as prioritizing candidate genes for experiments or generating hypotheses from transcriptomic data.

Key features include:

• Multi‑library support: Simultaneous enrichment across dozens of curated libraries—GO, KEGG, Reactome, MSigDB, GWAS Catalog, GTEx, DrugMatrix, and many more—so a single query can reveal functional, disease, tissue, or drug associations.
• Customizable library sets: Clients can request all libraries or a curated subset (e.g., popular GO terms, top pathways) via configuration flags, keeping responses relevant and efficient.
• Dedicated GO tool: A specialized endpoint for Gene Ontology Biological Process enrichment, making it trivial to retrieve the most biologically meaningful annotations.
• Statistical rigor: Automatic correction for multiple testing ensures only robust associations are returned, reducing noise in downstream AI reasoning.

Typical use cases span bioinformatics pipelines and interactive research assistants:

• A data scientist asks an LLM, “Which pathways are enriched in my differential expression list?” and receives a clean set of KEGG or Reactome hits ready for visualization.
• A clinician‑researcher integrates the server into a diagnostic assistant that cross‑references patient gene variants against disease ontologies and GWAS signals.
• A computational biologist builds a VS Code extension that, upon selecting a gene list, calls the MCP server and injects enriched pathway tables directly into the editor.

Integration is seamless: any MCP‑compatible client—Claude Desktop, Cursor, VS Code, or a custom web app—can add the Enrichr server via a simple configuration line. Once connected, the LLM can invoke tools like or , receive structured JSON, and weave the results into its responses. This tight coupling eliminates manual API calls, promotes reproducibility, and accelerates hypothesis generation in both research and clinical settings.