Add taxon constraints to display

[kltm]

From Chris: See quickgo for examples.

[kltm]

Links and discussions would be helpful.

[cmungall]

First we need to standardize the json payload for terminfo. For a mockup:
http://owltools.googlecode.com/svn/trunk/OWLTools-Core/src/test/resources/example-class.json

I will add taxon sntrs as examples to this

will get DavidOS to weight in

(should probably be its own issue)

[cmungall]

Depends on #58

[cmungall]

We could still deliver this without the json payload - just add another 2 fields for Golr. One for only_in_taxon, another for never_in_taxon. Each has a value from ncbi taxonomy ontology. We want the full {id,label,closure} quad for only_in_taxon. never_in_taxon is more subtle.

@hdietze and I would do the majority of the work here to populate the field in the golr loader. @kltm would just tweak a conf file in the amigo client so that these two fields would show up on the ontology view (if people want to facet by them - ask @dosumis as the ontology group). And just show the direct values for the two fields.

[kltm]

I've added this to the 2.2 milestone for when we climb into the Java stuff after Caltech.

[kltm]

Loader stuff, so now bumped to 2.3.

[cmungall]

Note: closing this one as a dupe: https://www.ebi.ac.uk/panda/jira/browse/GO-219

[paolaroncaglia]

Hi,

Is there any update on this please? I.e. displaying taxon constraints in AmiGO term pages. At present, they seem to be completely invisible. GO survey says ~36% of our users refer to AmiGO, so that's a shame if we can't convey that information to such a large portion of users...

I was looking into this as part of the taxonomy update paper, but it's really a broader issue.

Thanks!
Paola

[kltm]

Assuming the above works (and that's a big "if"), it will initially only be exposed for terms. We'll have to make a slightly more invasive change at the owltools loader level to get it exposed for annotations as well. Next load is likely for the weekend, but I may trigger sooner as I'd probably like to freeze before the meeting.

[cmungall]

Happy for this ticket to just be about terms, and to be able to display the taxon constrs on the term info page. Can open a new ticket for adding these into annotation documents but there isn't really a use case as the actual taxon in the association will be more specific than the constraint.

[kltm]

We'll be reverting here:

Yuuup--that's us. This was actually following up on Paola's request
yesterday. It should have been a gimme, but we had not tested it before
and were waiting for the weekend reload to see (an immediate failure was
a possibility, and we'd just restart after reverting if we had
problems). I then pushed it out of my mind, waiting for the Jenkins
report. Justin--thank you, you got very (un)lucky on this upstream grab.

Practically, this would probably mean then that the expected output for
those functions is not actually what the flexloader wants. I'll revert
here and wait until we have time to dig into owltools. 

@elserj

[kltm]

@hdietze I believe that the issue here is that getOnlyInTaxonLabels (returning a set) should be getOnlyInTaxonLabel (returning a string). I'd bet that the parser chokes on the fact that a cardinality-of-one function is returning a set instead of a string.

[kltm]

@elserj, I believe that this should be sufficient if you want to test, or you can keep what you have.

[kltm]

Okay, I've completely backed out for now--this cannot be done without a schema change: the id/label fields can have a cardinality of multi, so we need to add the map to the schema and populate it as well. This will take a little more work, so we're going to kick it for a bit.

We should have reverted everything now to the older "good" state. Older, but wiser.

Actions:

• change schema to make get getOnlyInTaxon and getOnlyInTaxonLabels multi
• add mapping fields, and use new @hdietze function above to populate it
• roll out new schema with changes

@elserj you should monitor this ticket as it will require a schema upgrade at your end as well when we make the jump.

[paolaroncaglia]

Hi all and thanks for trying anyway...

Just to have an idea (I don't mean to bug you, but I'd like to quote this plan in a publication on taxonomy representation in GO), what do you think a reasonable timeframe for this task would be, given resources? And would it be ok if I reminded you after the holidays? :-)

Thanks,
Paola

[kltm]

After the holidays is fine. It probably would not take that much effort to sit down and coordinate the code and schema upgrade, but there are a lot of low and medium effort tasks floating around that currently have priority. 😞 If there is a deadline or other goal to hit, that can help us order what needs to be done.

[paolaroncaglia]

Thanks Seth. Let's update after the holidays, and we can set a time plan then.

[cmungall]

Note that this ticket may be subsumed in the fix for Planteome/planteome-core#20

[cmungall]

Discussed on 2023-11-16 call

Now we are including the precomputed TCs in the released ontology (geneontology/go-ontology#19759) we can do this easily with a simple lookup on the json graph.

Note we can also easily add this to closures

amigo/metadata/ont-config.yaml

Lines 233 to 245 in c6a4840

	- id: only_in_taxon_closure
	description: "Only in taxon closure."
	display_name: Only in taxon (IDs)
	type: string
	cardinality: multi
	property: [getDummyStrings]
	- id: only_in_taxon_closure_label
	description: "Only in taxon label closure."
	display_name: Only in taxon
	type: string
	cardinality: multi
	property: [getDummyStrings]
	searchable: true

Change the dummy method to getRelationIDClosure(RO:0002162)

It would be nice to do this before the 10th birthday of this issue :-)

[kltm]

Okay, talking to @cmungall , when the next GO release goes live, I'll take a look to see if this data has propagated into the json graph and, if so, do some template render tweaks to expose it in the interface. V/easy.
If not there for some reason, I'll try and see what changes would be necessary.

[kltm]

Alas, due to issues with an API change at Zenodo, we are likely to miss the birthday :(

[kltm]

geneontology/pipeline#345

[kltm]

A bit of a hash in there, but there is information that can be extracted.
https://golr-staging.geneontology.io/solr/select?defType=edismax&qt=standard&indent=on&wt=json&rows=10&start=0&fl=*,score&facet=true&facet.mincount=1&facet.sort=count&json.nl=arrarr&facet.limit=25&hl=true&hl.simple.pre=%3Cem%20class=%22hilite%22%3E&hl.snippets=1000&fq=document_category:%22ontology_class%22&q=id:%22GO:0000165%22
Probably easiest crutch is to use the NCBITaxon namespace.
Noting that we don't have relation labels available there, but

sjcarbon@moiraine:~/local/src/git/go-ontology/src/ontology[master]$:) grep -oh -P "RO_[0-9]{7}" imports/go_taxon_constraints.owl | sort | uniq -c
    872 RO_0002160
    590 RO_0002161
   2564 RO_0002162

(Also noting that the icons have really rotted away and need fixing. Bleh.)

[kltm]

Missed the birthday a little, but now exposed in https://amigo.geneontology.org/amigo/term/GO:0000165

[pgaudet]

Great ! This is super useful.

However I wonder about the taxonomy page - when I click through 'Eukaryotes,' I get to this page, with 12 annotations - that seems surprising?

[kltm]

@pgaudet Yeah, I don't think that's "technically" wrong, as it's answering the question of "what annotations have the term NCBITaxon:2759 in their is-a/part-of closure". The answer seems to be a complete circle with CA3 pyramidal cell dendrite, so there is probably something to be explored on the ontology side with that (tagging @balhoff), but not necessarily a direct AmiGO question.
That said, on the AmiGO side, if this is confusing to users, there's no reason not to explore suppressing this (for now anyways) by either linking to the external NCBITaxon page (easy), removing the linkout (easy), or suppressing annotation searches in non-GO terms pages (doable, but spider sense says that that might not be what we want).

[pgaudet]

I think it would be much more appropriate to link to NCBI taxonomy.

suppressing annotation searches in non-GO terms pages

OK, I see now that following an extension to Uberon leads to the OLS page (for example, C1 field of hippocampus) here
https://amigo.geneontology.org/amigo/gene_product/UniProtKB:P05067, I didn't know this had been changed, but I propose we link out for external ontologies.

Thanks, Pascale

[kltm]

Not sure why our link generation is different here. I'll update.

[kltm]

The link is now in line with all other taxon links in AmiGO.