search

obophenotype / ncbitaxon

Build for NCBITaxon
BSD 3-Clause "New" or "Revised" License
27 stars 7 forks source link

Add definition for "strain" #107

Open DanBerrios opened 6 months ago

DanBerrios commented 6 months ago

Taxon doesn't generally have definitions, and that's appropriate for most of the concepts in the database. However, the ranks could at least be defined, no? We would like to point to some ontology-based concept definition for strain, and the one in OBI is limited to selectively maintained organisms.....

Britannica says: "Among domestic animals, a true-breeding, genetically pure line is usually called a strain."

If this issue has been raised before, let me know, or if it is more appropriate to define strain and other subspecies concepts in other OBO ontologies, please refer me to them...

matentzn commented 6 months ago

Cc @sabrinatoro we need some kind of general answer for these questions?

DanBerrios commented 5 months ago

This issue was discussed during OBI dev calls in Mar and Apr '24. Some notes:

'strain' rank in NCBITaxon is widely used, but had no definition and a bad identifier: http://purl.obolibrary.org/obo/NCBITaxon_strain
there is a "taxrank" ontology in OBO, which has a bunch of taxonomy ranks but does not have 'strain': http://obofoundry.org/ontology/taxrank.html
Population and Community Ontology has various 'collection of organism' terms http://purl.obolibrary.org/obo/PCO_0000000 but does not have 'strain'

Bjoern suggests this distinction:

NCBI Taxon for natural organism evolutionary classification
PCO for populations of organisms in the wild
OBI for inbred organisms for experiments
VBO has breeds of animals
FOODON need cultivars for agriculture...

Bjoern Peters: "Summarizing the conclusions: The general 'strain' concept is not in OBI's primary scope, so we will not create it here. It could be included in COB to coordinate the different classes mentioned above from different OBO ontologies. ..."

sabrinatoro commented 4 months ago

@matentzn: Apologies, this fell off my radar... In VBO, we decided to not define terms like strains/sub-breeds/varieties/etc because the community disagrees on what these terms mean, and it isn't always clear what the distinction is between these concepts. Instead, we consider terms referring to any of these concepts subclasses of a general breed concept which we broadly defined as "a group of animal which shares common characteristics such that they are considered an entity on their own". I hope that helps!

cmungall commented 4 months ago

Sorry for the slow response @DanBerrios

In fact the correct place to put definitions of taxrank is in the taxrank ontology (https://www.ebi.ac.uk/ols4/ontologies/taxrank).

We have a few separate issues for this:

Now, it gets a bit more complicated here. taxrank does not have "strain". It has "bio-variety" which may or may not be related, but if we want to use taxrank directly in the OBO NCBITaxon rendering then we need the exact same NCBI concept (imprecise as it might be) in taxrank, OR we go with a non-isomorphic mapping (#55)

DanBerrios commented 4 months ago

@sabrinatoro I agree there is disagreement on how to scope "strain" concept, but however it winds up being defined (it can be defined very inclusively), "strain" should not be limited to animals, so we can include virus strains, etc.
@cmungall I checked out the metaclasses slide presentation. I like the biolink:category option. If I interpreted the proposal correctly, then BALB/c (a mouse "strain") would be an instance of metaclass "strain" (? defined in some namespace, I guess TaxRank?) with biolink:category_of (? is that the proposed inverse property of biolink:category? ) -> NCBITaxon "Mus musculus musculus".... for example.... Is that the proposal?

DanBerrios commented 4 months ago

@sabrinatoro I agree there is disagreement on how to scope "strain" concept, but however it winds up being defined (it can be defined very inclusively), "strain" should not be limited to animals, so we can include virus strains, etc.
@cmungall I checked out the metaclasses slide presentation. I like the biolink:category option. If I interpreted the proposal correctly, then BALB/c (a mouse "strain") would be an instance of metaclass "strain" (? defined in some namespace, I guess TaxRank?) and with biolink:category_of (? is that the proposed inverse property of biolink:category? ) -> NCBITaxon "Mus musculus musculus".... for example.... Is that the proposal?