Figure 4.

LinkHub relational and RDF data models. (a) LinkHub relational model (b) An example RDF graph of two statements or triplets: (Q60996, gene_annotation, GO:0005634) and (Q60996, protein family, PF01603), which describe that there is a protein (Q60996) whose gene annotation is identified by GO: 0005634 and protein_family by PF01603. This also exemplifies using an RDF graph to connect multiple resources. Here, it connects UniProt, Gene Ontology, and Pfam. The detailed description associated with each identifier can be provided by the corresponding resource (the URL or URI can provide access to such detailed descriptions). (c) Illustrates how the key LinkHub relational tables identifier_types, identifiers, and mappings (reproduced above the RDF structure) are mapped to the corresponding RDF structure. The resulting RDF graph captures different types of object identifiers stored in different databases and the relations (or mappings) between these object identifiers. The mapping types are explicitly represented as RDF properties. (d) Shows how the rest of the LinkHub relational tables (reproduced above the RDF structure) map to the RDF structure. The resulting RDF graph captures the different web resources (which can be grouped) accessible by LinkHub. In addition, the graph captures information about which web resources accept which types of object identifiers, as well as exceptions.

Smith et al. BMC Bioinformatics 2007 8(Suppl 3):S5   doi:10.1186/1471-2105-8-S3-S5