search

OBOFoundry / COB

An experimental ontology containing key terms from Open Biological and Biomedical Ontologies (OBO)
https://obofoundry.github.io/COB
Creative Commons Zero v1.0 Universal
39 stars 8 forks source link

Replace site->immaterial entity in spreadsheet #11

Closed cmungall closed 5 years ago

cmungall commented 5 years ago

Rationale:

alanruttenberg commented 5 years ago

Spatial regions can't be part of material entities, but sites can.

cmungall commented 5 years ago

when would anyone in our domain (or any domain) ever need to use SRs?

alanruttenberg commented 5 years ago

I'm not sure how that is relevant? If no one uses SR then we don't include the term SR.

I'm going on the principle that if we mean site, then we should say site. If the label isn't good, then we can add an alternative label. Or lobby for a change in BFO's label.

Whether it matters, and to what extent, depends on usage.

Immaterial entities include boundaries as well, BTW

cmungall commented 5 years ago

@dosumis any preferences? I personally prefer least commitment. Because I don't understand SRs I don't know for sure whether some of the IEs we use may later turn out to be SRs, which would break things if we overcommit. Also, BFO is not closed. Maybe there will be other kinds of IEs which are neither SR or sites, and our entities may turn out to be in a mix.

If we do commit to site then really we should do this across the board, and we should rename all terms like 'immaterial anatomical entity' to 'anatomical site' and deepen the existing subClassOf axioms.

bpeters42 commented 5 years ago

I prefer to keep it to 'sites'. 'immaterial entity' as a label sucks, as most people would assume that to include roles, functions, information content and the like. Worse, it would allow for spatial regions, which I would like to test if excluding them completely causes any problems with OBO ontologies. If it does, I am open to generalizing to immaterial entities. Testing these things is what our work is for, so we don't have to make a decision, and should be open to either.

As a side note, I would love to allow for 2d/1d sites, which would make me pretty confident that we would not need any other immaterial_spatial entities for OBO

On Tue, Feb 12, 2019 at 2:12 PM Chris Mungall notifications@github.com wrote:

@dosumis https://github.com/dosumis any preferences? I personally prefer least commitment. Because I don't understand SRs I don't know for sure whether some of the IEs we use may later turn out to be SRs, which would break things if we overcommit. Also, BFO is not closed. Maybe there will be other kinds of IEs which are neither SR or sites, and our entities may turn out to be in a mix.

If we do commit to site then really we should do this across the board, and we should rename all terms like 'immaterial anatomical entity' to 'anatomical site' and deepen the existing subClassOf axioms.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/OBOFoundry/Experimental-OBO-Core/issues/11#issuecomment-462957722, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9IjYjNIhPoAi_ca55K1aSkpXeeBGtks5vMzxigaJpZM4ath5I .

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

cmungall commented 5 years ago

I just remembered boundary is under IM. This is a useful concept. So ignoring ugly labels or over/under commitment, we would either need to bring in IM as a grouping, or implement this pattern https://github.com/BFO-ontology/BFO/issues/221 to get power of owl reasoning

bpeters42 commented 5 years ago

'site' is the lost 3-dimensional sibling of the zero/one/two dimensional 'continuant fiat boundaries' in bfo. There is no apparent reason for their different treatment in bfo. Personally, I would love to have a grouping term for them (and I thought 'site' is just fine, as it doesn't imply dimensionality in common usage, and as a site transitions changes shape, it can go from 3d to 2d, as in the site defined to be between a hydraulic press piston and the base support ( https://www.youtube.com/watch?v=os9qYgJ6Nw8)

For many sites, (like the peptide binding site on an MHC molecule), I really don't want to have to specifiy if it is 2 or 3 dimensional. this also touches on Darren's work for PRO.

So after rambling so much, I guess I am saying that I hate it but I agree that we should generalize to immaterial entity for now. I would at least like to see if we can explicitly catch and exclude use of spatial regions though, which on our past inspection was always incorrect for OBO ontologies.

On Tue, Feb 12, 2019 at 5:38 PM Chris Mungall notifications@github.com wrote:

I just remembered boundary is under IM. This is a useful concept. So ignoring ugly labels or over/under commitment, we would either need to bring in IM as a grouping, or implement this pattern BFO-ontology/BFO#221 https://github.com/BFO-ontology/BFO/issues/221 to get power of owl reasoning

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/OBOFoundry/Experimental-OBO-Core/issues/11#issuecomment-463018588, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9IsS-QT2FQ4By23enbMl7iJvOCBeTks5vM2ysgaJpZM4ath5I .

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

dosumis commented 5 years ago

I prefer to keep it to 'sites'. 'immaterial entity' as a label sucks, as most people would assume that to include roles, functions, information content and the like.

I'm not aware of that confusion. Isn't it only ontologists who think of functions roles and information content as entities? Maybe it depends on context? I think it's highly unlikely that anyone seeing the term 'immaterial entity' in an anatomy ontology would assume it refers to any of these things.

'Site' feels like a slightly odd term - in my mind it immediately begs the question - what's special about a space/immaterial entity that makes it a site? It's also loaded, e.g. in the context of proteins, people are likely to think of active-sites. In the context of anatomy at least, if we want something more specific than 'immaterial entity' then I'd go for 'cavity'.

Question - how deep will OBO-core be? are we going for a single layer of intermediate terms between BFO and everything else, or will it have some nested structure. e.g. if cavity is too specific (implies material boundary which may not the case with site?), then could OBO-core have

site . cavity ?

alanruttenberg commented 5 years ago

@dosumis What makes site different from spatial region is that a site is anchored to a material entity. Consider two people with infections, walking towards each other. You can take the people's position as relative to a spatial region, but then the sites of the infections occupy different spatial regions at different times.

If boundaries and sites are both needed, and SR not, then I suggest leaving sites and adding boundaries. If someone can think of a unifying parent, then add that too - it doesn't need to be in BFO since it's completely defined by the union of the two classes.

bpeters42 commented 5 years ago

Agree with what Alan is saying. But am also fine with 'immaterial entity' for now - the confusion I am referring to won't affect the people currently working on the core, and we can search for a unifying parent of site + boundary later.

On Wed, Feb 13, 2019 at 5:28 AM Alan Ruttenberg notifications@github.com wrote:

@dosumis https://github.com/dosumis What makes site different from spatial region is that a site is anchored to a material entity. Consider two people with infections, walking towards each other. You can take the people's position as relative to a spatial region, but then the sites of the infections occupy different spatial regions at different times.

If boundaries and sites are both needed, and SR not, then I suggest leaving sites and adding boundaries. If someone can think of a unifying parent, then add that too - it doesn't need to be in BFO since it's completely defined by the union of the two classes.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/OBOFoundry/Experimental-OBO-Core/issues/11#issuecomment-463199080, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9Il5wcWP_hJbCoXKEdlNitkzzJA6bks5vNBMMgaJpZM4ath5I .

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

dosumis commented 5 years ago

@dosumis What makes site different from spatial region is that a site is anchored to a material entity.

I get that this is (was?) the intent of the BFO term, but I thought with OBO core we were looking for intermediate level terms with intuitive names. If that is the case, perhaps we're still too far up in abstraction here, and we should instead be looking for more specific terms to expose in OBO-core. I suspect that cavity could cover all sites we need to refer to in anatomy, and would also be useful for other domains. If I understand correctly, a site is broader than a cavity - for example covering the portion of atmosphere above Greater London. Maybe we should retain site in OBO core for that purpose, preferring cavity in all other cases?

cmungall commented 5 years ago

Alternate proposal: obo-core only brings in IC (perhaps giving it a nicer name, but we shouldn't use the extent of ugliness of the BFO name dictate the modeling)

Caring about subdivisions of the IC becomes the business of domain ontologies.

Note we'll still have D+R axioms that are more specific, but this will be true unless we include 85% of BFO. I think we just get around this with the BFON pattern.

cmungall commented 5 years ago

I think this can be closed now since we aren't using the spreadsheet anymore?