Migrate CHEMINF to BFO 2?

[Janna Hastings]

Dear fellow CHEMINF developers,

for eNanoMapper I need to integrate a large number of ontologies under a common upper level. Having two versions of BFO in different ontologies is impossible. Therefore, I'd like to migrate CHEMINF to BFO 2 (currently it depends on BFO 1.1.)

I've already done this locally and it involves the following changes:

- removing the direct import of IAO and instead importing the 'iao-main.owl' and other dependent classes available  at http://information-artifact-ontology.googlecode.com/svn/releases/2012-07-11-BFO2/ in cheminf-external.owl (along with importing BFO 2)

- removing copied BFO classes from cheminf-external.owl

- editing a few references in cheminf.owl to BFO 1.1. classes to now use the equivalent BFO 2 classes instead.

Are there any objections to my committing these changes?

Thanks, Janna

[Egon Willighagen]

+1

> --
> You received this message because you are subscribed to the Google Groups
> "cheminf-ontology" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to cheminf-ontolo...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers
ORCID: 0000-0001-7542-0286

[Michel Dumontier]

Hi Janna,

  Can you show me where the OBO community had agreed to use BFO2? In particular, there seems to be substantial disagreement with respect to the time sensitive relations. 

m

--

[Janna Hastings]

Hi Michel,

the OBO community has indeed not agreed, certainly not on the relations. Still, we can exclude the relations by using the "classes-only" version of BFO 2. We anyway import the RO relations.

Cheers, Janna

[Michel Dumontier]

Hi Janna, 

  for the benefit of CHEMINF users, can you articulate the differences between BFO and BFO2? what is the advantage for CHEMINF? which ontologies are moving ahead with BFO2 despite the lack of consensus, and what does this mean for ontologies that don't do so?

m

[Janna Hastings]

Hi Michel,

For the differences between BFO 1.1 and BFO 2 see the release notes: http://purl.obolibrary.org/obo/bfo/2012-07-20/ReleaseNotes

Aside from the plumbing (URL pattern change to purl.obolibrary.org/obo/BFO_xxxx and use of rdfs:label) there are no differences that affect CHEMINF. All the BFO classes that CHEMINF used e.g. quality, specifically dependent continuant, are still there unchanged.

I want to migrate only to have consistency across a large group of ontologies that I need to integrate. OBI, for example, already moved.

If this doesn't sound like a good enough reason, then just say no, I'll make a local copy of CHEMINF and use that instead.

Cheers, Janna

[Michel Dumontier]

Hi Janna,

On Tue, Jul 22, 2014 at 8:16 AM, Janna Hastings
<janna.h...@gmail.com> wrote:
> Hi Michel,
>

> For the differences between BFO 1.1 and BFO 2 see the release notes:
> http://purl.obolibrary.org/obo/bfo/2012-07-20/ReleaseNotes

ok thanks. that's helpful to understand the differences.

i'm a little bit worried about the disclaimer : "the OWL version of
BFO will undergo further evolution and can therefore not be considered
stable"

should we be using something that isn't stable or adopted? It's too
bad that they didn't just opt to reuse the existing identifiers to
ensure forward compatibility.

> Aside from the plumbing (URL pattern change to
> purl.obolibrary.org/obo/BFO_xxxx and use of rdfs:label) there are no
> differences that affect CHEMINF. All the BFO classes that CHEMINF used e.g.
> quality, specifically dependent continuant, are still there unchanged.

would you be able to point us to your development version to evaluate?

> I want to migrate only to have consistency across a large group of
> ontologies that I need to integrate. OBI, for example, already moved.

can you list the set of ontologies that i) you need to integrate, and
ii) have already moved?

> If this doesn't sound like a good enough reason, then just say no, I'll make
> a local copy of CHEMINF and use that instead.

i'm not adverse to it ... i just want to make sure that we understand
the differences, the risks, and the benefits.

m.

[Egon Willighagen]

Michel, all,

On Tue, Jul 22, 2014 at 7:03 PM, Michel Dumontier
<michel.d...@gmail.com> wrote:
> i'm a little bit worried about the disclaimer : "the OWL version of
> BFO will undergo further evolution and can therefore not be considered
> stable"
>
> should we be using something that isn't stable or adopted? It's too
> bad that they didn't just opt to reuse the existing identifiers to
> ensure forward compatibility.

Depends on what they define as stable. If the doesn't mean changing
IDs, completely overhauling the design... that is, if future changes
doesn't change the semantics of what we use, it should be fine, or?

Egon

[Michel Dumontier]

On Tue, Jul 22, 2014 at 10:09 AM, Egon Willighagen
<egon.wil...@gmail.com> wrote:
> Michel, all,
>
> On Tue, Jul 22, 2014 at 7:03 PM, Michel Dumontier
> <michel.d...@gmail.com> wrote:
>> i'm a little bit worried about the disclaimer : "the OWL version of
>> BFO will undergo further evolution and can therefore not be considered
>> stable"
>>
>> should we be using something that isn't stable or adopted? It's too
>> bad that they didn't just opt to reuse the existing identifiers to
>> ensure forward compatibility.
>
> Depends on what they define as stable. If the doesn't mean changing
> IDs, completely overhauling the design... that is, if future changes
> doesn't change the semantics of what we use, it should be fine, or?

We'd obviously like that to be the case, as we believed when we
adopted BFO in the first place. but what if the BFO2 ontology contains
axioms that, if mapped to BFO1, and to CHEMINF, proves to be
incompatible?

m.

[Egon Willighagen]

On Tue, Jul 22, 2014 at 7:13 PM, Michel Dumontier
<michel.d...@gmail.com> wrote:
> We'd obviously like that to be the case, as we believed when we
> adopted BFO in the first place. but what if the BFO2 ontology contains
> axioms that, if mapped to BFO1, and to CHEMINF, proves to be
> incompatible?

Yeah, introducing "unit tests" for such things is one thing we want to
automate in eNanoMapper.

But I guess the same holds for having a combined ontology with both
BFO1 and BFO2 as I think Janna is referring to as the current
situation for us... (as some ontologies we want to use use different
versions...)

Choices are not final, nor these lists complete in anyway, but:

http://bioportal.bioontology.org/projects/eNanoMapper
https://jenm.bigcat.maastrichtuniversity.nl/view/Ontology/

(to give some idea...)

The second has some initial unit tests:
https://jenm.bigcat.maastrichtuniversity.nl/view/Ontology/job/Ontology%20Testing/

[Michel Dumontier]

On Tue, Jul 22, 2014 at 10:20 AM, Egon Willighagen
<egon.wil...@gmail.com> wrote:
> On Tue, Jul 22, 2014 at 7:13 PM, Michel Dumontier
> <michel.d...@gmail.com> wrote:
>> We'd obviously like that to be the case, as we believed when we
>> adopted BFO in the first place. but what if the BFO2 ontology contains
>> axioms that, if mapped to BFO1, and to CHEMINF, proves to be
>> incompatible?
>
> Yeah, introducing "unit tests" for such things is one thing we want to
> automate in eNanoMapper.
>
> But I guess the same holds for having a combined ontology with both
> BFO1 and BFO2 as I think Janna is referring to as the current
> situation for us... (as some ontologies we want to use use different
> versions...)
>
> Choices are not final, nor these lists complete in anyway, but:
>
> http://bioportal.bioontology.org/projects/eNanoMapper
> https://jenm.bigcat.maastrichtuniversity.nl/view/Ontology/

sorry

> (to give some idea...)
>
> The second has some initial unit tests:
> https://jenm.bigcat.maastrichtuniversity.nl/view/Ontology/job/Ontology%20Testing/

cool. this is a good idea Egon :)

please include a consistency test for CHEMINF. I would also suggest to
setup a test for the proposed CHEMINF-BFO2 dependency.

m.

> Egon
>
>
> --
> E.L. Willighagen
> Department of Bioinformatics - BiGCaT
> Maastricht University (http://www.bigcat.unimaas.nl/)
> Homepage: http://egonw.github.com/
> LinkedIn: http://se.linkedin.com/in/egonw
> Blog: http://chem-bla-ics.blogspot.com/
> PubList: http://www.citeulike.org/user/egonw/tag/papers
> ORCID: 0000-0001-7542-0286
>

[Janna Hastings]

Hi Michel and Egon,

the simplest way for you to see the proposed changes without my committing them is to overwrite your cheminf.owl and cheminf-external.owl with the attached versions. This uses the classes-only version of BFO 2 so we can stick with the old (RO) relations (minimizing the impact of the change).

Cheers, Janna