Provide standard place for union taxa

[cmungall]

carried from EnvironmentOntology/envo#1046

GO mints union taxa in a fake namesapce, NCBITaxon_Union. See https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-11-530

PCO is making union taxa in the PCO namespace PopulationAndCommunityOntology/pco#58

we should have a single place for union taxa. This should not be injectedinto the ncbitaxon namespace as ncbi don't make these. however, releases should be synced with ncbitaxon. we should use the go union pattern

[kaiiam]

cc @ramonawalls.

@cmungall would we want the new namespace to be hosted within NCBITaxon ontology or perhaps a different one that imports from it?

[ramonawalls]

Agree that something associated with NCBItaxon is a better home for these classes than PCO. I need to research alga to make sure we can actually define it as a union class, though. It may not include all members of the taxa it covers. There are also tricky classes, like 'microbe', which can't be defined taxonomically. Would those move here, or stay in PCO?

[cmungall]

new namespace, but coordinated with ncbitaxon releases

good point that not all can be defined by union. we can always just do subclasses if it can't be defined. But we should be careful that we have a maintenance plan. If we add a concept for microbe, what is our plan for ensuring reasonable completeness of classification? Is the gain for having a named concept worth it if it's completely unreliable? is it really that useful to have a concept for microbe given no one can even agree what one is?

Unfashionable opinion: sometimes it's OK not to have a class for a concept in an ontology. What I often do here is a born-obsolete class, so that people doing a search will find a term and some explanation of why it wasn't included as a non-deprecated class.

[balhoff]

@ddooley I feel like this discussion might be relevant to something you were asking about a while back in NCI-Thesaurus/thesaurus-obo-edition#51.

[kaiiam]

Unfashionable opinion: sometimes it's OK not to have a class for a concept in an ontology.

I think it's still important to have purls for commonly used organismal classes or groupings, e.g., microbe algae, even if we don't have a full owl constraint of who all is included. People will still want to annotated data with such terms.

[cmungall]

Another example of an ontology minting paraphyletic classes: http://purl.obolibrary.org/obo/FOODON_03411222 (fish)

[kaiiam]

From a food science perspective having a general concept for fish is needed.