search

obi-ontology / obi

The Ontology for Biomedical Investigations
http://obi-ontology.org
Creative Commons Attribution 4.0 International
75 stars 26 forks source link

chmo status #1005

Open rvita opened 5 years ago

rvita commented 5 years ago

Should we should contact CHMO about their registry status? Colin Batchelor is the responsible party for Chemical Methods Ontology. There is some possibility that he may want to retire it and move it to OBI.

StroemPhi commented 2 years ago

I don't think that Colin wants to retire CHMO and in the NFDI4CHem project context we reuse and contribute to CHMO. There is however definitely a need to harmonize CHMO and OBI. (see for example https://github.com/rsc-ontologies/rsc-cmo/issues/22, https://github.com/rsc-ontologies/rsc-cmo/issues/14 or https://github.com/rsc-ontologies/rsc-cmo/issues/13).

I thus suggest to have a call where the overlap between CHMO and OBI can be discussed.

bpeters42 commented 2 years ago

Colin had explicitly said he wanted to retire CHMO.

On Wed, Nov 23, 2022 at 4:30 AM Philip Strömert @.***> wrote:

I don't think that Colin wants to retire CHMO and in the NFDI4CHem project context we reuse and contribute to CHMO. There is however definitely a need to harmonize CHMO and OBI. (see for example rsc-ontologies/rsc-cmo#22 https://github.com/rsc-ontologies/rsc-cmo/issues/22, rsc-ontologies/rsc-cmo#14 https://github.com/rsc-ontologies/rsc-cmo/issues/14 or rsc-ontologies/rsc-cmo#13 https://github.com/rsc-ontologies/rsc-cmo/issues/13).

I thus suggest to have a call where the overlap between CHMO and OBI can be discussed.

— Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/1005#issuecomment-1324989049, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADJX2IXFTMY4A32Z4VDNMO3WJYE55ANCNFSM4G762FDA . You are receiving this because you were assigned.Message ID: @.***>

-- Bjoern Peters Professor La Jolla Institute for Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

bpeters42 commented 2 years ago

I just went back over emails, and what I wrote is not entirely correct. In emails from 2016, Colin agreed that CHMO terms should be 'adopted in OBI'. Which is not completely the same as retiring. The tracker discussion is here, the emails were off tracker: https://sourceforge.net/p/obi/obi-terms/794/

On Wed, Nov 23, 2022 at 5:12 AM Bjoern Peters @.***> wrote:

Colin had explicitly said he wanted to retire CHMO.

On Wed, Nov 23, 2022 at 4:30 AM Philip Strömert @.***> wrote:

I don't think that Colin wants to retire CHMO and in the NFDI4CHem project context we reuse and contribute to CHMO. There is however definitely a need to harmonize CHMO and OBI. (see for example rsc-ontologies/rsc-cmo#22 https://github.com/rsc-ontologies/rsc-cmo/issues/22, rsc-ontologies/rsc-cmo#14 https://github.com/rsc-ontologies/rsc-cmo/issues/14 or rsc-ontologies/rsc-cmo#13 https://github.com/rsc-ontologies/rsc-cmo/issues/13).

I thus suggest to have a call where the overlap between CHMO and OBI can be discussed.

— Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/1005#issuecomment-1324989049, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADJX2IXFTMY4A32Z4VDNMO3WJYE55ANCNFSM4G762FDA . You are receiving this because you were assigned.Message ID: @.***>

-- Bjoern Peters Professor La Jolla Institute for Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

-- Bjoern Peters Professor La Jolla Institute for Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

StroemPhi commented 2 years ago

I guess we need to get @batchelorc involved in this discussion 😉. So what do you, Colin, think of having some way of coordinating with OBI what should be defined there and what in CHMO?

cmungall commented 1 year ago

Here is a radical proposal:

OBI adopts a strict "no unsanctioned external subclassing". If you want a new term or new branch, add it to OBI.

By the same token, there should be procedures in place for rapid incorporation of new terms and for granting "ownerships" of branches or subsets of OBI to communities who could be trusted to manage that part of the ontology, review PRs.

This is more extreme than the no injection policy I have been advocating in OBO for a while (e.g. https://github.com/rsc-ontologies/rsc-cmo/issues/22). But this will avoid so many problems. I think we have observed this cycle with many ontologies many times:

  1. community A wants new terms in ontology B
  2. community A perceives (fairly or not) that B is unable or too slow to adopt new terms
  3. community A creates a new ontology A, that subclasses leaf or leafy nodes from B
  4. everyone else is confused about which terms to use from which ontologies or where to request new terms
  5. there is no mechanism in place in ontology B to say "subclasses of this term should go in A", and over time B's editors start unwittingly duplicating content in A
  6. the initial enthusiasm that led to development of A wanes, new terms are not added
  7. new design patterns and practices adopted by B are not propagated to A, leading the merged product to look more frankenstein-like
  8. the wider community is even more confused about where to request new terms that are scope for both A and B
  9. everyone just uses MESH instead
StroemPhi commented 1 year ago

While I definitely see your outlined benefits from the "no unsanctioned external subclassing" proposal, this would cripple a true modular approach of developing ontologies imho. Wouldn't OBI thereby be doomed to transgress its scope borders? I think the main problem is missing good documentation that aids the users and developers of ontologies to really know the agreed upon and to be adhered to patterns and proper places of terms/branches.

Having said that, I'd like to be more pragmatic here and only focus on the CHMO and OBI overlap, trying to tackle one problem at a time. Could a pragmatic approach be:

  1. CHMO will be ODKized, to have a better maintainability and standardization wrt import modules (BFO, RO, IAO, OBI, ...)
  2. CHMO will try to obsolete all terms that already have an eqivalent in OBI and file issues whenever there is a conflict to be resolved,
  3. as CHMO "is intended to be complementary to the Ontology for Biomedical Investigations (OBI)", it will be ok to subclass "leafy" terms from OBI to represent concepts that are more clearly in the chemical domain, as CHMO now adheres to OBI design patterns and would at best get an OBI dev review on their PRs to make sure this remains so?
StroemPhi commented 1 year ago

To be more concrete (and adressing 2.), let's look at chmo:mass spectrometry and obi:mass spectrometry assay. Just by looking at the children, siblings and parent in CHMO, I would assume CHMO to be the better "home" for mass spectrometry.