apaonto URLs don't resolve

[cmungall]

https://bioregistry.io/registry/apaonto

The expansion for apaonto is the very suspicious:

http://ontology.apa.org/apaonto/termsonlyOUT%20(5).owl#

This looks like a mistake in setting up the new entities plugin in Protege, with it default to an autogenerated Mac filename

And not unexpectedly, the APA PURLs don't resolve

[cthoyt]

What’s the fix here? That’s what’s in the ontology. Even if it doesn’t actually mean anything, this is what you get when using URIs sometimes…

Update: just emailed the 3 people listed on BioPortal to see if they could comment:

• Alvin Walker, awalker@apa.org (responded, said he will take a look at this issue)
• alexander garcia, alexgarciac@gmail.com
• Ian Galloway, igalloway@apa.org (bounced, email is dead)

[cthoyt]

@alexgarciac can you comment?

[cmungall]

I can think of a few general fixes for cases like this:

1. only include bioportal prefixes that pass some minimal URI resolution and namespace sanity check
2. make all bioportal-sourced prefixes that don't have precedent resolve to purl.bioontologies.org
3. don't auto-ingest bioportal at all (but of course people would be free to register on a case-by-case basis)

I favor 2.

cc @graybeal

[cthoyt]

only include bioportal prefixes that pass some minimal URI resolution and namespace sanity check

Entries that only showed up in BioPortal but nowhere else are only added to the Bioregistry manually if they are needed to standardize xrefs from relatively higher quality resources, like OBO Foundry ontologies. IIRC one of the disease/phenotype ontologies makes xrefs to apaonto.

make all bioportal-sourced prefixes that don't have precedent resolve to purl.bioontologies.org

So for your suggestion on 2, it would be https://bioportal.bioontology.org/ontologies/APAONTO/?p=classes&conceptid=http%3A%2F%2Fontology.apa.org%2Fapaonto%2FtermsonlyOUT%2520%285%29.owl%23$1 which takes the actual URLs inside and puts them at the end of a string constructed the bioportal prefix? Maybe something we could build more generally on top of https://github.com/cthoyt/ontoportal-client since it would require introspecting on terms and some guesswork. However, since there's only a small number of prefixes in the Bioregistry that only map to BioPortal, this can probably be handled manually.

don't auto-ingest bioportal at all (but of course people would be free to register on a case-by-case basis)

BioPortal will NEVER be fully auto-ingested in the Bioregistry. There is simply way too much relatively low quality content. This shouldn't come as a surprise, though, since it's written in several places describing the alignment workflow and the metaregistry. Again, that doesn't mean that we can't add stuff on a case-by-case basis

[graybeal]

In the end that seems like an illegal IRI (not the BioPortal one, the original one).

This is likely irrelevant, but: Normally you don't have to encode an IRI to get it to resolve in BioPortal, but in this case https://bioportal.bioontology.org/ontologies/APAONTO/?p=classes&conceptid=http%3A%2F%2Fontology.apa.org%2Fapaonto%2FtermsonlyOUT%2520%285%29.owl%23Abandonment the actual identifier looks like http://ontology.apa.org/apaonto/termsonlyOUT%20(5).owl#Abandonment which reflects an embedded space and parens, not sure either of those are acceptable in IRIs. Anyway, I can't build a shorter IRI that resolves, it appears.