Consolidate NIDM OWL Files with NIDM Repo and Issues

[nicholsn]

This proposal is to migrate the NIDM Vocabulary files (e.g., nidm-results.owl) from the NI-O repository into the NIDM repository to:

• simplify issue management and avoid confusion between repos
• avoid scope creep between ontology and vocabulary development

Open Questions:

• do we also migrate the dicom terms?
• where should the owl files go?
• is there interest in a longer term goal of harmonizing with other ontology efforts (e.g., OBI)?

[nicholst]

+1

[cmaumet]

+1

[nicholsn]

Anyone else have thoughts on this? If not, I'll open a pull request to move things over once #87 is resolved.

[satra]

i think the longer term goal should be kept in mind - but we should also keep other neuroimaging ontologies in mind (e.g., neurolog - http://bioportal.bioontology.org/ontologies/ONL-MR-DA). however, in the context of vocabularies, we should definitely be looking into these ontologies so that we don't replicate effort (@khelm - there might be quite a bit of overlap between the dicom work and ontoneurolog).

+1 for migrating dicom terms (and harmonizing with neurolog - especially given that several definitions are lacking in neurolog). these terms can always be slurped up by neurolex, but i think the flexibility of manipulating them is easier in this repo.

[dbkeator]

I like keeping all the ontology files/terms stuff with the NIDM data modeling stuff. I don't think we need to maintain a separate repo for it. Regarding the dicom term work, well, I'm on the fence about that. The NIDM repo makes sense (to me) being a repo for the data models and all associated files such as terminologies, ontologies, etc. If we expect to create a data model of some DICOM files (maybe dataset descriptors or experimental hierarchies) then it makes sense to have the terms/ontology in the same repo. If the DICOM terms work stands on its own (i.e. doesn't have an associated NIDM data model hook) then it might be better to live in it's own dedicated repo...

[cmaumet]

While we are currently more focused on defining vocabularies for the NIDM part, my understanding was that our long-term goal would be to create a proper ontology that would live in the ni-o repository.

Maybe the work on DICOM terms, which is closer to an ontology could stay in ni-o?

[nicholsn]

@khelm is traveling currently, so I think we can table the dicom term discussion until he is back in a few weeks. IMHO, it would be nice for that work to be marketed to the broader biomedical imaging community, as it is domain agnostic.

Aside from dicom, are we good with moving over the nidm vocabulary files to their respective component directories in incf-nidash/nidm?

[dbkeator]

@cmaumet I guess one question is whether the proper ontology will in fact just be an ontology of the terms relating specifically to the data models or will it be a broader ontology of things related to concepts outside the scope of the data models. If it's the former then I could see it living with the data models, if the latter then it makes sense to have a NI-O repo....

Yes, I guess that's what I'm thinking with respect to the DICOM terms...

[satra]

@nicholsn - our dicom work has been quite focused on MR based imaging as opposed to all the things that dicom supports. i'm not sure @khelm or our group has the resources to support the broader engagement. while it would be great to engage others to maintain it, i think the current dicom vocabularies are extremely relevant to our work as we chain provenance together. even for openfmri we are planning on creating dicom exports at least for the datasets where that information is available (ds113 for example), so dicom terms are as relevant to our current needs as other terms.

re: moving to specific locations - will we have a list of files to import for a given application? i'm thinking alongside import statements in languages. would there be an import all somewhere?

[nicholsn]

@satra, no disagreements there - I'm not proposing we do any community engagement currently. I just think the work is relevant beyond neuroimaging and would like to see other communities reuse what we've done. That being said, we could stick to the same model as with the other nidm vocabulary pieces and move over a file called something like nidm-dicom.owl that just captures the terms. Does that sound good?

re: imports, we can use owl:import in a file called nidm.owl that will pull in each of the component vocabularies into the top level nidm namespace. For a given application, I think they can create a separate "application vocabulary owl file" that only imports the pieces they need. I can work up examples of this once we agree to move forward.

[satra]

@nicholsn - sounds good.

[cmaumet]

We discussed this on NIDASH call on June 30th.

[JessicaTurner]

You mean June 30?

[cmaumet]

Thank you @JessicaTurner, this is now corrected.

[nicholsn]

from the call it sounded like this will work. once we merge #87, I'll create a PR to move these over.

[nicholsn]

Created PR #103 to merge over the vocabulary files and will close this now, so the conversation can continue over there.