Using large language models to create a knowledge graph of related data at EBI – with a searchable AI-based interface.

EMBL-EBI has and continues to collect a huge amount of data relevant to human health, from the scale of chemicals (ChEMBL), proteins (UniProt), genomes (ENSEMBL), andpathways (Reactome) through to genomic studies across populations (GWAS Catalog, EVA, EGA, ArrayExpress), and environmental data such as biomes in wastewater and soil (MGnify). While there are many use cases for each dataset individually, linking multiple datasets together holistically will allow the full research potential to be realised. Using large language models, this project sets out to create a knowledge graph of related data at the EBI with a searchable AI-based interface.