Closed kltm closed 8 years ago
@cmungall had this list suggested:
Mammalia
Vertebrata
Fungi
Metazoa
Viridiplantae
Eukaryota
Archaea
Bacteria
Note the taxon field here still shows the full closure:
which is impossible to browse to leaf taxa
It should be consistent with this: http://amigo2.berkeleybop.org/amigo/term/GO:0022008
To clarify, it looks like a different issue--we switched one off of taxon_closure_label to just taxon_label. Open a new issue on switching all filters to taxon_label (easy config fix).
@cmungall temporary usability step has been separated as #250.
Is there a blocker on me here? Should we add annotations to ncbitaxon to indicate what is in the slim?
Or maybe this should just be a yaml config list? The planteome folks will likely want more granularity in the plant taxa used.
As discussed, tagging @hdietze as well; apparently #209 got the owltools part done? For the loader, we'll have two fields, names TBD: one for the full taxon closure and one for the slim.
Miracle of miracles, we already have this in the schema from a long time ago: "taxon_closure_subset" "taxon_closure_subset_label" "taxon_closure_subset_map" We may want to have another widget personality for this?
http://purl.obolibrary.org/obo/go/extensions/go-taxon-subsets.owl
Not yet imported into go-plus
404: Do you know how long these usually take to percolate out?
every 30 mins. Next in at 2:40pm
Should we test this as a separate ontology in the load, or wait until it is imported?
Last question resolved by @hdietze .
At some point we will want to make the ordering (at least partially) configurable rather than data-driven. This is too much for 2.4 but making a note here as the approach for the slim may overlap the approach for this.
Use "amigo_group_subset" for new option provided by @hdietze.
(Optional) flag will be, for example:
--solr-taxon-subset-name amigo_group_subset
Something is not right in our local tests--using the new code path seems to prevent the generation of anything taxon related in the annotation docs, or at least for getting to Solr (as a lot of the taxon info still goes by). That said, the generated ontology documents seem to be spot on.
To ensure that nothing unknown has change, removing
--solr-taxon-subset-name amigo_group_subset
and
http://purl.obolibrary.org/obo/go/extensions/go-taxon-subsets.owl
and retrying.
Even after rolling back the two items above, I'm still not getting anything loaded besides taxon. amigo_group_subset no longer there, but the same 124 ncbi terms.
Well, this leads us back to either something subtle changed in owltools or something subtle at my end. I have been changing a large amount of metadata recently, but am a little lost on what that mechanism could be. A quicker test might just be going back a bit in owltools and seeing if there is an improvement.
Ever weirder. Re-trying with owlcollab/owltools@7f70c3b, I still get empties. I'm going to dig into metadata changes that I've made and see if there are any leads there.
@hdietze / @cmungall I've walked back to owlcollab/owltools@9c72e14 and still no luck, so maybe not owltools. I've also undone a bunch of stuff at the amigo end (although I'm not sure how any of that could prevent owltools from doing annotation loading--amigo stuff should only have an effect on schema generation and flexloader AFAICR), and no luck. I'll poke around some more in there.
I guess another avenue to pursue would be any changes to the ontologies? Anything gone on in there recently?
Let's work on some junit unit/integration tests for this. I'm not following exactly what the issue is but seems it should be caught upstream.
More clear examination here in the morning with @hdietze. The immediate problems with losing the taxon and evidence closures have been traced back to ontology selection in the load, this has been cleared away. We're left with no subset closure yet, so we (he) will look toward the owltools end.
correction the subset is named: amigo_grouping_subset
--solr-taxon-subset-name amigo_grouping_subset
Now just need to switch all displays over to the new one.
The code for the improvement was finished in #209, now all that is needed is a subset with the name "model_slim", either as a separate ontology on in the taxon slim. Changed to over here to reflect that it is now a data issue that can be taken care of at individual sites rather than waiting on code. For GO, we'll aim at the next release.