Species specificity
From OBO Foundry
Species and Taxon specificity
The OBO Foundry covers the biology of many different varied organisms. This poses particular challenges for ontology development. Ontologies are sometimes taxon-centric, in that their scope is limited to a particular set of organism types; sometimes they are taxon-neutral in that there is no limit on the range of organisms covered
Some examples:
- Gene Ontology (GO) : taxon-neutral
- Foundational Model of Anatomy : taxon-specific - Human
- Plant Ontology (PO) : taxon-species - All plants (neutral within plants)
- Zebrafish anatomical ontology (ZFA) : taxon-specific - Danio rerio (zebrafish) - being generalised to teleosts
Deprecation of sensu
Many OBO ontologies have adoped the sensu design principle previously used in GO to indicate taxon-specific terms. However, this practice is now discouraged in GO in favour of explicit differentia, and the recommendation is that this is followed in other OBO ontologies.
Previously, GO would have had an is_a hierarchy like this:
cell wall cell wall (sensu bacteria)
The definition would have been along the lines of "a cell wall, as in but not restricted to the taxon Bacteria"
This is problematic for a number of reasons: it is imprecise (due to the qualifier about "not restricted to"), and it places a dependence on the species taxonomy.
The recommended practice is to always provide a differentium (discriminating characteristic) when creating a new is_a child node that is appropriate for the ontology and not species-taxon related. For the GO cellular component ontology, this would be a structural differentium. The new GO hierarchy is as follows:
cell wall A protective structure outside the cytoplasmic membrane composed of peptidoglycan, a molecule made up of a glycan (sugar) backbone of repetitively alternating N-acetylglucosamine and N-acetylmuramic acid with short, attached, cross-linked peptide chains containing unusual amino acids; also called murein. As in, but not restricted to, the taxon Bacteria (Bacteria, ncbi_taxonomy_id:2). cell wall
The definition is: A protective structure outside the cytoplasmic membrane composed of peptidoglycan, a molecule made up of a glycan (sugar) backbone of repetitively alternating N-acetylglucosamine and N-acetylmuramic acid with short, attached, cross-linked peptide chains containing unusual amino acids; also called murein
Note that the previous "sensu" term is retained as a synonym. The ID is preserved, as the intended meaning is preserved. There is still a note about Bacteria in the definition "gloss":
A protective structure outside the cytoplasmic membrane composed of peptidoglycan, a molecule made up of a glycan (sugar) backbone of repetitively alternating N-acetylglucosamine and N-acetylmuramic acid with short, attached, cross-linked peptide chains containing unusual amino acids; also called murein. As in, but not restricted to, the taxon Bacteria (Bacteria, ncbi_taxonomy_id:2).
It can still be beneficial to include knowledge of associations between terms and species taxons. The proposal is to do this as annotations, outside the ontology proper, using Waclaw Kusnierczyk's relations (in press).
Note that the GO editorial style guide has not been updated and still reflects the old recommendations: http://www.geneontology.org/GO.usage.shtml#using
Material relating to this can be found here:
Anatomical Ontologies
Creation of ontologies that are valid for multiple species can be challenging. Often we want to make statements like
X part_of Y (but only in T)
The old pattern would have been to create a new term "X sensu T", and to add two links:
XsT is_a X XsT part_of Y
Occasionally this would propagate, leading to
XsT is_a X YsT is_a Y XsT part_of Y
The new pattern is to replace the sensu terms with structural differentia. In many cases, this may just require a name and definition change
