Skip to main content
Fig. 1 | Journal of Biomedical Semantics

Fig. 1

From: Learning from biomedical linked data to suggest valid pharmacogenes

Fig. 1

Overview of the type of entities and relationships considered and their origin. Entities are of three distinct types: Gene, Phenotype and Drug. Gene–Phenotype relationships are coming from ClinVar and DisGeNET, Phenotype–Drug relationships from SIDER and Medi-Span, Gene–Drug relationships from DrugBank. In addition, we included gene and drug entities from PharmGKB to enable building the training and test sets. Equivalence mappings are defined between entities of the same type but of different origin. In addition to entity–entity relationships, we consider some attributes that are specific to entities, such as the ATC class of drug that is a drug attribute. Naming of different parts of the data (e.g., G–P links, gene attributes) is used later in the step of formatting of the linked data. The detailed schema of the data is provided Fig. 2

Back to article page