organismal groupings for EBI/JGI GOLD microbiome terms

[kaiiam]

Cross link to #672

Within the EBI/JGI GOLD's root > Host-associated hierarchy there are a many organismal groupings which are not monophyletic clades, and cannot be described by an existing NCBITaxon semantic. Examples include Reptilia, Protists, Protozoa, Spiralia, fish and green algae.

@diatomsRcool @ramonawalls as I would like to have terms to link the new microbiome terms to, I am wondering which Ontology would be best suited host such organismal groupings, perhaps ECOCORE or PCO?

Perhaps under the organism hierarchy in ECOCORE as they already have some classes there such as algae, prey as well as the three domains of life imported from NCBITaxon?

[kaiiam]

There are also symbiosis-related terms which include Ectosymbionts, Extracellular symbionts, Intracellular endosymbionts (including primary and secondary), as well as Epibionts. @pbuttigieg and I agreed that these should likely go to ECOCORE.

[cmungall]

For GO taxon constraints we create non-monophyletic unions, e.g prok = arch OR bact

http://purl.obolibrary.org/obo/go/imports/go-taxon-groupings.owl

The specific owl module may not fit your needs but you could use the same approach. Define your classes using OWL DL constructs. Reason over these using hermit in advance to get your subclass hierarchy - don't assert.

[cmungall]

We could release these alongside NCBITaxonm but it wouldn't be appropriate to have NCBITaxon IDs thouggh of course

[kaiiam]

@cmungall Ideally I would like to have resolvable purls to an ontology hosted on Ontobee, would it be possible to have defined classes which have such axioms or do you think this kind of approach is better?

[pbuttigieg]

The approach @cmungall suggests will likely get done faster, but having terms in ECOCORE / PCO would be more sustainable in the long run (unless EBI or another user means something very specific when they use such a grouping).

[diatomsRcool]

I would like to work with @ramonawalls to figure out if ecocore or pco is more appropriate.

[cmungall]

Great. And I think suggestions are complementary. I would suggest the same methodology as GO but housing them in an appropriate ontology with resolvable PURLs.

…

On Fri, May 31, 2019 at 9:18 AM diatomsRcool ***@***.***> wrote: I would like to work with @ramonawalls <https://github.com/ramonawalls> to figure out if ecocore or pco is more appropriate. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#781?email_source=notifications&email_token=AAAMMOMGRFXK44M4BVIYSSTPYFFSRA5CNFSM4HP35JS2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWVWBXQ#issuecomment-497770718>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAAMMOIQQK5Q237TJWK3VGTPYFFSRANCNFSM4HP35JSQ> .

[kaiiam]

@cmungall just to be clear my understanding of how you created the go-taxon-groupings.owl using DL constructs was that you did the following:

To create the Prokaryota class, in protege you go to the DL query tab

Then in the DL query box you add what becomes the equivalent to axiom:

NCBITaxon_2 or NCBITaxon_2157 Where NCBITaxon_2 is Bacteria and NCBITaxon_2157 is Archaea.

Then click the Add to ontology button? Then you run the HermiT reasoner and it will figure out the subclass hierarchy? Is this correct?

If @ramonawalls and @diatomsRcool decide to go with regular classes, couldn't we achieve the same effect by creating a new class and giving it an equivalence axiom for example NCBITaxon_2 or NCBITaxon_2157 for a prokaryotes class?

[cmungall]

You could do it that way but the more conventional way as just as you would add any other class with a defining equivalence axiom, through the class descriptions view. And you would have the ontology loaded so you'd see labels not IRI fragments I'm not sure I follow your distinction re regular classes..

…

[kaiiam]

I was looking into DL construction classes, and I thought that they were a different type of owl class more like owl lite, and that you would have to make it a different way in Protege but if the distinction isn't important then it probably doesn't matter.

We'd just create classes and give them equivalence axioms with OR's. Correct?

[cmungall]

Yep - some may require a combination of OR, NOT + AND. Eg "fish" in common parlance = vertebrata and not tetrapoda

…

On Tue, Jun 4, 2019 at 10:01 AM Kai Blumberg ***@***.***> wrote: I was looking into DL construction classes, and I thought that they were a different type of owl class more like owl lite, and that you would have to make it a different way in Protege but if the distinction isn't important then it probably doesn't matter. We'd just create classes and give them equivalence axioms with OR's. Correct? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#781?email_source=notifications&email_token=AAAMMOISCCZQ34YKXAM4BJTPYZYRLA5CNFSM4HP35JS2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODW4VDHY#issuecomment-498684319>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAAMMOL2TOPNUFCPPVF73DTPYZYRLANCNFSM4HP35JSQ> .

[kaiiam]

Great thanks @cmungall. I'd be happy to make these contributions to ECOCORE and or PCO once it's settled where they should go.