SGD links to complexes do not work

[pgaudet]

SGD annotates complexes with SGD identifiers, for example http://amigo.geneontology.org/amigo/gene_product/SGD:S000218003
However the links we provide from AmiGO are not working
https://www.yeastgenome.org/locus/S000218003

@suzialeksander can you please check why these links are not resolving?

Thanks, Pascale

[suzialeksander]

So, the /locus/SGDID only works with locus pages. For a complex, the URL is /complex/SGDID, so yeastgenome.org/complex/S000218003. Can you split the SGDIDs by type?

[kltm]

@suzialeksander GO policy has been that CURIEs do not contain information that needs to be parsed in the internal ID section. That would be that either SGD supplies a redirect to the right location for the general case or that the ID space is split so that the namespace is used to get things to resolve correctly. There is a similar discussion centered around #659 (comment) ; TAIR and WB have gone through similar questions.

[cmungall]

options:

1. create a SGD.COMPLEX namespace, analogous to what you did for pathways. bioregistry already has both:
   • https://bioregistry.io/registry/sgd.pathways
   • https://bioregistry.io/registry/sgd
   • so here would would have SGD.COMPLEX:S000218003 and this is the CURIE we would use in GO
2. have the link go to a https://www.yeastgenome.org/search?q=S000218003&is_quick=true (though the complex doesn't seem to show up in the search?)

[pgaudet]

Note that the SGD complexes ID to CPX mappings should also be in the SGD GPI file
GOA cannot parse these either.

[srengel]

we are requesting please to use the #2 option:

have the link go to a https://www.yeastgenome.org/search?q=S000218003&is_quick=true

[pgaudet]

@kltm This is still not working as expected - ie the links are not as requested by @srengel https://www.yeastgenome.org/search?q=S000218003&is_quick=true

[kltm]

Okay, current SGD metadata is:

go-site/metadata/db-xrefs.yaml

Lines 2372 to 2415 in fe54521

	- database: SGD
	name: Saccharomyces Genome Database
	synonyms:
	- SGDID
	rdf_uri_prefix: https://identifiers.org/sgd
	generic_urls:
	- http://www.yeastgenome.org/
	entity_types:
	- type_name: gene
	type_id: SO:0000704
	id_syntax: S[0-9]{9}
	url_syntax: https://www.yeastgenome.org/locus/[example_id]
	example_id: SGD:S000006169
	example_url: https://www.yeastgenome.org/locus/S000006169
	- database: SGD_LOCUS
	name: Saccharomyces Genome Database
	generic_urls:
	- http://www.yeastgenome.org/
	entity_types:
	- type_name: entity
	type_id: BET:0000000
	url_syntax: https://www.yeastgenome.org/search?q=[example_id]&is_quick=true
	example_id: SGD_LOCUS:GAL4
	example_url: https://www.yeastgenome.org/search?q=S000006169&is_quick=true
	- database: SGD_REF
	name: Saccharomyces Genome Database
	generic_urls:
	- http://www.yeastgenome.org/
	entity_types:
	- type_name: entity
	type_id: BET:0000000
	url_syntax: http://www.yeastgenome.org/reference/[example_id]/overview
	example_id: SGD_REF:S000049602
	example_url: http://www.yeastgenome.org/reference/S000049602/overview
	- database: SGD_PWY
	name: Saccharomyces Genome Database YeastPathways
	generic_urls:
	- https://pathway.yeastgenome.org/
	entity_types:
	- type_name: entity
	type_id: BET:0000000
	url_syntax: https://www.yeastgenome.org/locus/[example_id]
	example_id: SGD_PWY:CYSTEINE-SYN2-PWY
	example_url: https://pathway.yeastgenome.org/YEAST/NEW-IMAGE?type=NIL&object=SERSYN-PWY

In there, right now, there is a separate SGD_LOCUS namespace, which has the "search" link. The base SGD namespace ("SGD") uses the direct locus link.

Is the ask here that the base namespace also uses the search, or that SGD is going to start using more split namespaces. If the former, it would be as simple as adding the search link to that base namespace and doing an update.

[suzialeksander]

I think these are switched, where logically the base namespace should use the search, the LOCUS namespace should use the direct locus link.

adding the search link to that base namespace and doing an update.

This is what we want. It would fix the complex links correct?

[kltm]

@suzialeksander Correct.

[kltm]

@suzialeksander For example: #2150

[pgaudet]

OK on amigo-staging

[kltm]

Will close when checked on production AmiGO

[kltm]

@pgaudet This looks like it's working using the original example. Please close if confirmed.