Enhancing Human Proteome Exploration with MCP and Neo4j

Session Track: AI Engineering

Session Time: 08:30 pm - 09:00 pm November 6

Session description

Exploring the vast human proteome is a major scientific challenge. Although LLMs are powerful, they often provide inaccurate or hallucinated answers for such a complex domain, limiting their use for serious research. In this session, Juan Aguirre will showcase a powerful, real-world application of Neo4j and Model Context Protocol (MCP). He will demonstrate how he implemented this pattern on a human proteome knowledge graph to ground LLM responses and enable reliable scientific exploration. This talk will show a practical implementation of the GraphRAG pattern, using MCP as the structured approach for retrieving factual, real-time context. You will see how to model biological data in a graph, implement MCP, and apply GraphRAG techniques to enhance LLM accuracy in complex domains.