GrEBI
Aggregates biomedical knowledge graphs using ontologies and LLM embeddings to facilitate integrative queries for humans and machines.
Acerca de
GrEBI (Graphs@EBI) is a high-performance computing (HPC) pipeline designed to integrate knowledge graphs from diverse biomedical resources such as EMBL-EBI, MONARCH Initiative, ROBOKOP, and Ubergraph. By leveraging ontologies and LLM embeddings, GrEBI aims to simplify complex, integrative queries that span multiple data sources, overcoming the limitations of single-resource REST APIs. The pipeline produces two main outputs: periodically updated materialized query results as gzipped CSV files, and downloadable Neo4j database exports. It is currently applied in significant projects including the International Mouse Phenotyping Consortium (IMPC) knowledge graph and the EMBL Human Ecosystems Transversal Theme (HETT) ExposomeKG.
Características Principales
- HPC-driven aggregation of diverse biomedical knowledge graphs.
- Integrates LLM embeddings and ontologies for enhanced knowledge representation.
- Provides materialized query results in CSV format for easy data loading and analysis.
- Offers downloadable Neo4j database exports for local or HPC-based querying.
- Facilitates cross-resource integrative queries via an API/MCP server.
- 2 GitHub stars
Casos de Uso
- Analyzing large-scale aggregated knowledge graph data using readily available materialized query results.
- Conducting integrative queries across disparate biomedical knowledge graphs to uncover novel insights.
- Deploying and querying comprehensive biomedical knowledge graphs on local systems or HPC clusters for custom research.
