Fig. 2
From: Learning from biomedical linked data to suggest valid pharmacogenes
Schema of the pharmacogenomic linked data selected for this study. Entities are related to either Genes, Phenotypes (or Diseases) or Drugs. We artificially enriched the data with an additional type of entity: gene–drug pairs. These entities link exactly one gene and one drug and are the nodes of the graph we classify either as associated or not associated from a PGx point of view, to valid candidate pharmacogenes. For mapping purposes, we added to our dataset Gene references from UniProt and Drug references from PubChem. Because part of Medi-Span and part of PharmGKB data are protected, we restricted the online access to the data