Open vanaukenk opened 3 years ago
creation-date, modification-date, and import-date are now in the ShEx.
@vanaukenk I plan to export creation-date
and import-date
as annotations; modification date is what we're already exporting in the dedicated column. Does this sound right?
@balhoff Sorry for the late reply.
Yes, creation-date
and import-date
would be included as annotations in the Annotation Properties field of the GPAD, and modification date will be used to populate the dedicated Date field of the GPAD file.
Sorry for the duplicate ticket. I didn't have this in the MGI imports project.
Looks like we need to have a quick meeting about the OWL model. For GPAD export, Minerva retrieves these dates off of evidence instances. They don't seem to be in that position in the generated OWL.
Discussed with @ukemi: the creation and import dates should be retrieved from the relation axioms, rather than the evidence nodes.
I think this makes sense since they are not associated with the evidence.
Fix for this is in dev
branch now.
I still don't see creation-date in the properties of the GPAD annotations that are derived directly from models that were made using the Noctua interface. I do see them in the imported annotations. You can distinguish the models by the model ID. Ones made directly are a hash-like string and ones that are imported have an MGI identifier.
Directly made: MGI MGI:1095407 located_in GO:0005905 PMID:16923969 ECO:0000314 20201217 MGI part_of(GO:0005886) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:5fce9b7300000367|model-state=production
Imported:
MGI MGI:1888520 acts_upstream_of_or_within GO:0043983 PMID:16339075 ECO:0000315 20101018 MGI occurs_in(CL:0000352) contributor=https://orcid.org/0000-0001-7476-6306|**creation-date=2010-10-18**|model-state=production|noctua-model-id=gomodel:MGI_MGI_1888520|**import-date=2021-08-09**
MGI MGI:1888520 enables GO:0005515 PMID:16109376 ECO:0000353 UniProtKB:O60583 20060804 MGI creation-date=2006-08-04|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693|noctua-model-id=gomodel:MGI_MGI_1888520|**import-date=2021-08-09**
MGI MGI:1888520 enables GO:0005515 PMID:16109376 ECO:0000353 UniProtKB:P50750 20060804 MGI creation-date=2006-08-04|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693|noctua-model-id=gomodel:MGI_MGI_1888520|**import-date=2021-08-09**
The directly made models don't have that information within them. I don't know if that's the responsibility of Noctua or Minerva, but we should create a separate issue if we need to add that somewhere.
That's what I suspected, but good to have you confirm. We can do a workaround wrt the load, but we do need to consider this as we move forward. @vanaukenk
I'm going to give the go-ahead with the merge and suggest we open another ticket for the noctua-generated models. It really is a separate issue.
From 2021-09-28 MOD imports call:
For now, suppress exported 'import date' annotation property.
We can re-visit this decision in the future, if it becomes useful to groups/users.
@vanaukenk the update is implemented; I moved the card to 'needs testing'.
Thanks @balhoff. I will review this when @dustine32 turns the crank to generate the new MGI models. My target is to start the new rounds of review on Monday.
Tested on the latest GPAD from Dustin. It appears to have worked. OK to send to production.
Although creation dates are being incorporated into the file, the question has arisen wrt what we mean by a creation date. Is it the date the first assertion between a GP and a term was made (this is how the GPAD is currently being generated) or is it the date an original/conventional annotation was modified? In conventional annotation every assertion and evidence is considered a separate annotation. In the new world is an annotation a GP-term relationship supported by evidence?
From 2021-11-02 MOD imports call:
As part of the import code, @dustine32 will now populate evidence with the creation_date value in each line of the annotation properties field of the imported GPAD2.0 file.
We'll see how this looks in the graph editor and then decide if we need to do any work on the display.
For GPAD1.2 output, @balhoff will then query the evidence to output the correct creation_date annotation property value.
With https://github.com/biolink/ontobio/issues/599 I've generated MGI import models with the created
annotations on Evidence individuals. An example currently loaded to noctua-dev is:
http://noctua-dev.berkeleybop.org/editor/graph/gomodel:MGI_MGI_3028590
All Evidence individuals in this model should have a <http://purl.org/dc/terms/created>
annotation:
@balhoff I think this should be enough for you to query out the created
value for the minerva GPAD export? Let me know if not.
@dustine32 Please let me know when the new WB models are ready, as well. These were nearly ready to go to production Noctua, but I'll check the export GPAD again once they've been regenerated. Thx.
@vanaukenk You probably already know this, but the updated models are now available on noctua-dev. An example model shows dc:created
now in place:
In our latest round of imports, it seems that everything is working correctly. @vanaukenk ?
However, we probably shouldn't push this out to the public production files yet since we are only accommodating for it in our test environment.
This is working fine for the WB imports and latest GPAD export files, although we don't require it for the roundtrip back into WB.
@ukemi - we'll leave this on noctua-dev until you give the thumbs up for MGI.
Note that since this is a change to the Noctua-derived GPAD, we should probably announce this to the consortium so any groups who pick up those files are forewarned.
I just double-checked on the latest file and things seem to be working ok. I still think we should test that file with our production software to see if anything breaks.
Can this be closed?
I think so.
I am just looking at the GPADs on snapshot and I don't see the date tags:
MGI MGI:2387642 enables GO:0008009 GO_REF:0000008 ECO:0000266 UniProtKB:Q6UXB2 20160819 MGI part_of(GO:0007204),part_of(GO:0070098) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:5716c41300000277|model-state=production
So I am reopening this.
@ukemi that model http://noctua.geneontology.org/editor/graph/gomodel:5716c41300000277 has no creation dates within it.
Ha. Trying to do too many things at once:
MGI MGI:2135272 acts_upstream_of_or_within GO:0001843 PMID:12499390 ECO:0000315 MGI:MGI:1857642 20040831 MGI contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:MGI_MGI_2135272|model-state=production
mutant phenotype evidence used in manual assertion contributor: https://orcid.org/0000-0001-7476-6306 date: 2004-08-31 dc:created: 2003-03-04 providedBy: http://informatics.jax.org 🔗 source: Disruption of scribble (Scrb1) causes severe neural tube defects in the circletail mouse with: MGI:MGI:1857642
@ukemi it looks like we had a mixup in communication:
I disabled this until we decided to make it public. But then this plan slipped my mind. Should I go ahead and enable?
Since we just closed out the Imports project and this is by no means a show stopper, I put it in the new PLANNING Project. Let's hold off until we can prioritize.
For the MOD imports project, we want to export creation-date and modification-date in the annotation properties field (field 12).
This depends on work being done here:
https://github.com/geneontology/go-shapes/pull/264 https://github.com/geneontology/go-shapes/pull/263
Further specifications on exactly how these two annotation properties will be populated will come once the above is resolved.