Define intended behavior for obsolete ontologies in ontology browsers

[cmungall]

An example of an obsolete ontology is SAO, which was merged into GO

on the OBO site we retain the original info, but have a prominent banner indicating status and replacement:

http://obofoundry.org/ontology/sao

However, browsers act differently

SAO appears as any other ontology in OntoBee:

• http://www.ontobee.org/ontology/SAO

It is a 404 in OLS

• https://www.ebi.ac.uk/ols/ontologies/sao

In bioportal it appears as any other ontology, but there is a note that says "Retired"

• https://bioportal.bioontology.org/ontologies/SAO

In aberowl it appears as any other ontology

• http://aber-owl.net/ontology/SAO/#/

While there is an argument for retaining some trace of an obsolete ontology for historic purposes, or for gap analysis etc, in the majority of cases users should be discouraged from using an obsolete ontology. There is a danger ontology annotators will suggest use of IDs from the obsoleted ontology rather than the replacement one, or that users will waste time annotating manually from an obsolete ontology.

I propose:

• browsers SHOULD omit obsolete ontologies from their index
• browsers MAY include obsolete ontologies in their index, provided:
  • the obsoletion status MUST be clearly marked for human consumers in both the main ontology page and individual term pages
  • the obsoletion status MUST clearly marked in programmatic API payloads, and obsolete ontologies SHOULD be masked by default from operations such as NER, and only included if the client passes a flag requesting inclusion

cc @graybeal @yongqunh @ebispot @henrietteharmse @leechuck