OBI asserts a GO:molecular function to be a subclass of GO:biological process

[balhoff]

(sorry this was submitted prematurely)

OBI contains these axioms:

• GO:sequence-specific DNA binding SubClassOf OBI:binding
• OBI:binding SubClassOf GO:biological_process

However in GO sequence-specific DNA binding is an indirect subclass of molecular_function. molecular_function and biological_process are disjoint. So this creates problems using OBI with GO.

I think it is the same situation with core promoter binding.

[bpeters42]

We would appreciate help from the GO community how to do this better. Hopefully @cmungall can chime in. The basic problem has been that OBI originally took the view that GO:'molecular function was a kind of BFO:function. But we were then made aware that GO treats molecular function as a type of BFO:process. It has not been clear to us what constitutes the difference between GO:molecular function vs. GO:biological process. OBI:binding was created when the only BFO:process in GO we were aware of was GO:'biological process'. And we wanted to describe the processes that occur during an assay that are being measured. It seems we can now refer to those as 'biological process or molecular function'.

OBI:binding was created as a placeholder that we would be happy to replace. There are editor notes in that regard on the term. And we we should now explore replacing OBI:binding with GO:binding (GO_0005488), which seems to work for all current children of OBI:binding.

However, originally we intended for OBI:binding to cover other binding processes, such as cell:cell interactions, which I suspect GO would put under 'biological process'. Minimally the label 'binding' seems very general for GO if it only applies to 'molecular function', however that is defined.

[balhoff]

@bpeters42 I agree the distinction between GO:molecular function and GO:biological process is unclear. We discussed this on the GO editors call today. GO does take the view that molecular functions are parts of biological process. Perhaps the solution for OBI would be to simply use BFO:process to group these terms instead of the GO:biological process. Or, perhaps the union you suggested. Regarding 'binding', I agree it is a pretty broad label; the GO definition is very specific.

@cmungall or @ukemi please add if I've left something out.

[cmungall]

Good summary

Perhaps GO should relabel it's binding to 'molecular binding' or somesuch - or at least have an obofoundry unique label property on it

[bpeters42]

So if I understand correctly, the hierarchy is: bfo:process GO:molecular function GO:binding (--> consider relabeling: GO:molecular binding) GO:biological process OBI:planned process OBI:assay OBI:binding assay (--> consider redefining: OBI:molecular binding assay) Correct? We can implement the way we import the GO terms in OBI then. And this is actually a perfect example of coordination to get the OBO-core right... - Bjoern

…

On Mon, Sep 10, 2018 at 12:31 PM Chris Mungall ***@***.***> wrote: Good summary Perhaps GO should relabel it's binding to 'molecular binding' or somesuch - or at least have an obofoundry unique label property on it — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#963 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ANN9IpOSqC0Ggk1mz40mUVVNS2OEsq2Lks5uZr4agaJpZM4WcBEZ> .

-- 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]

That's correct. I think we can even go one step further and assume all siblings are disjoint here (but not exhaustive)

[jamesaoverton]

I believe that this was addressed in #936 and the change has been released. Please reopen if I'm mistaken.