search

FAIRsharing / domain-ontology

A project supporting the DRAO application ontology, a hierarchy of specific research domains and descriptors which imports subsets of terms from over 50 publicly-available ontologies.
Other
1 stars 1 forks source link

Refactor EDAM Protocol and OBI Protocol #77

Closed allysonlister closed 3 years ago

allysonlister commented 3 years ago

When moving the children of the EDAM Report class, I noticed that we have both EDAM Protocol and OBI Protocol. I'll find out how many of each are in FAIRsharing before I recommend which we keep, but at the moment I think keeping the OBI term would be better due to its richer hierarchy within DRAO.

Originally posted by @allysonlister in https://github.com/FAIRsharing/domain-ontology/issues/71#issuecomment-773431894

allysonlister commented 3 years ago

@ramGranell has kindly let me know that 73 records have the EDAM term http://edamontology.org/data_2531, and 16 have http://purl.obolibrary.org/obo/OBI_0000272.

I am happy to write up the change more formally, but which class would you like to keep @delphinedauga and @Drosophilic?

I'd probably still suggest retaining the OBI term. Thanks!

Drosophilic commented 3 years ago

Hi @allysonlister - I vote for the EDAM term as it (to my mind) a better definition. The OBI term definition seems less clear.

delphinedauga commented 3 years ago

Hi both ! -> I like the Edam definition but the concept of reproducibility is missing (contrary to the OBI def) (Edam has also a small error in the def, about is duplicated "collection of information about about how a scientific ". Is there a way to notify them of this typing error?) -> Not really comfortable with the child_of "report" (Edam), as a report implies that the action (experiment) should have been done before, which is the contrary of a protocol. The definition of "plan specification" from OBI is more adapted. -> For these reasons, I prefer the OBI definition

allysonlister commented 3 years ago

@delphinedauga and @Drosophilic and I have spoken about this and decided to go with OBI Protocol. The following is a summary of our discussion.

Why Refactor Protocol

While working on #71 we realized that Protocol was present twice.

Both terms are visible in FAIRsharing.

Below are the two locations (highlighted) of the two infection terms in the current DRAO hierarchy. protocol

Why are you suggesting this change?

As this is an automatically-generated hierarchy, it is inevitable that these kinds of concept duplications will occur, and should be resolved as they are discovered.

Mapping

We have agreed to the following:

Old IRI New IRI Note
http://edamontology.org/data_2531 http://purl.obolibrary.org/obo/OBI_0000272 The EDAM term is not required, as we already have an OBI equivalent already in use.

Hierarchy

Decision: Align with OBI.

Reasoning: The EDAM definition is good, but the concept of reproducibility is missing (contrary to the OBI def). The position of the OBI protocol as a child of "plan specification" is preferred.

I'll move ahead with these changes.

allysonlister commented 3 years ago

The updates I've just performed show the following changes. Please note Report is mentioned because I noticed with this update that I had accidentally retained one outstanding annotation request within development/DRAO-ontofox-annotation.txt, which I have now fixed (#71 please take note).

Ontology comparison

Left

Right

Ontology imports

Ontology annotations

Gene report http://edamontology.org/data_0916

Removed

Added

Protocol http://edamontology.org/data_2531

Removed

data_2048 http://edamontology.org/data_2048

Removed

allysonlister commented 3 years ago

All curation has been performed within FAIRsharing to use the OBI term instead.

allysonlister commented 3 years ago

@delphinedauga has checked this modification.