should 'nucleate cell' be abandoned as unsustainable?

[GoogleCodeExporter]

The class graph under nucleate cell is very incomplete.  To make it complete we 
need either lots of assertion of multiple classification, which we know is not 
sustainable, or we need to add had_part some nucleus to most cells in the 
ontology.  This second option is not much more sustainable than the first. 

 Is it worth the hassle?  Should we just ditch the class?

COMMENTS PLEASE

Original issue reported on code.google.com by dosu...@gmail.com on 20 Sep 2013 at 4:25

[paolaroncaglia]

The class graph under nucleate cell is very incomplete. To make it complete we need either lots of assertion of multiple classification, which we know is not sustainable, or we need to add had_part some nucleus to most cells in the ontology. This second option is not much more sustainable than the first.

Is it worth the hassle? Should we just ditch the class?

COMMENTS PLEASE

Original issue reported on code.google.com by dosu...@gmail.com on 20 Sep 2013 at 4:25

Here's the current graph. Possibly less incomplete than in 2013, though surely not 100% complete. However, some of the subclasses may be potentially useful. ...Do we have a label for "can of worms"? :-)

[paolaroncaglia]

Originally opened by @dosumis .

[addiehl]

'nucleate cell' is one of the terms from the original CL that is of course problematic, since some cell types are named or defined with reference to their having one or more nuclei, or by their lack of a nucleus. On one hand, in order to have a proper accounting of nucleated cells, we would have to add appropriate axioms to all of them. On the other hand, it is mostly harmless to have the hierarchy as is, since the lack of a positive statement about the presence of a nucleus does not really matter for most cell types, since other defining criteria are more important.

At some point in the last few years, I did make an effort to define hematopoietic cells in a way that referenced their absence or presence of a nucleus, because these cell types seem to have the most variation in this attribute or are referenced in ways that refer to their 'nucleus number' (trying to avoid saying 'nuclear number') or even the morphology of their nucleus.

Options I would consider: 1) close this issue and not worry too much about this. Folks can add ('bearer of' some nucleated) to their favorite cell types and nobody will be too bothered. 2) obsolete 'nucleate cell' but keep its more interesting children 'single nucleate cell' and 'multinucleate cell'. Also keep 'anucleate cell', which is an interesting grouping class.

I can live with either option. It's the exceptions that are more interesting, and option 2 preserves the ability to talk about them if we decide that the 'nucleate cell' is mostly a pointless distinction that theoretically requires too much bookkeeping. -- Alex

[dosumis]

I favour option 2. I think it's generally a bad thing to have grouping classes that will always lack most of the subclasses anyone would expect them to have. I'd also remove 'single nucleate cell' as this suffers from exactly the same issue.

[paolaroncaglia]

Thanks @addiehl and @dosumis for your feedback! Playing devil's advocate, according to Ontobee 'nucleate cell' is used by 17 ontologies, including Mondo, HPO, uPHENO, CLO... similarly for 'single nucleate cell'. So either convert this ticket into an Epic, or go for Alex' option 1?

[dosumis]

I think the OntoBee report is just an indication of import, not actual direct usage - but just because some subClass is used.

If that's going to stop us obsoleting, we'll never get rid of terms like this.

Looking a bit further.

no sign of its direct usage in HPO here: https://ols.monarchinitiative.org/ontologies/hp/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FCL_0002242

Actually - it turns out that if you click through to the individual reports of usage in OntoBee you can see direct usage. This => just two examples:

CLO: http://www.ontobee.org/ontology/CLO?iri=http://purl.obolibrary.org/obo/CL_0002242 The legacy version of uPheno: http://www.ontobee.org/ontology/UPHENO?iri=http://purl.obolibrary.org/obo/CL_0002242 - I don't think we care about this (@matentzn - any opinion?)

The Epic here is I think to set up an SOP for obsoletion. I suggest we model this on GO - with announcement going out to a community list that we encourage known users to subscribe to. We can also directly contact any ontologies directly using the term - assessed via individual usage reports on ontoBee, although we might be able to automate reports in future. We should add this to the agenda for our next call.

For this ticket I suggest we contact CLO warning of obsoletion and obsolete if no objection from them, or after one month if no reply.

[addiehl]

I have mixed feelings about 'single-nucleate cell'. I think the term has been used to group terms like 'mononuclear cell', which are leukocytes specifically defined by the presence of a single non-segmented nucleus, and terms like 'nucleated thromboblast', where being nucleated is an exceptional feature relative to standard 'thromboblasts', ie platelets, found in most mammals. But of course, it has been applied poorly, since interesting cells like 'nucleate erythrocyte' are not grouped here due to a looser equivalence axiom.

[paolaroncaglia]

Also see https://github.com/obophenotype/cell-ontology/issues/561.