Stream: genomics
Topic: URI for HPO CodeSystem
Alejandro Metke (Oct 16 2018 at 23:46):
I just noticed that the URI for HPO in this spreadsheet is http://hl7.org/fhir/sid/hpo. We have been using the URI and version that are in the OWL version of HPO (url: http://purl.obolibrary.org/obo/hp.owl - version: http://purl.obolibrary.org/obo/hp/releases/2018-06-13/hp.owl). Given that the most straightforward way of creating the FHIR code system is through an OWL to FHIR transformation, I think it will make things easier to keep the identifiers already being used in the OWL version. Also, the people already using HPO will be familiar with these. Thoughts?
Grahame Grieve (Oct 16 2018 at 23:51):
they're not always the same thing in principle. e.g. the LOINC RDF identifiers are a transform away from LOINC codes in FHIR, and there's some value there
Alejandro Metke (Oct 17 2018 at 00:08):
Sure, but LOINC is not primarily authored in OWL, so it doesn't really have a primary, well-known identifier. When dealing with HPO and any other code systems that are primarily authored in OWL, there is already an existing identifier so I don't see much value in creating an additional one.
Andrew Patterson (Oct 17 2018 at 00:14):
In this particular spreadsheet they are listed as coming from the V2 LRI - but I don't see the URI defined in that document at all.. @Patrick Werner ?
Julian Sass (Oct 17 2018 at 07:46):
@Andrew Patterson the code systems come from the LRI, the URIs don't. We started to define those URIs, where there are only OIDs at the moment. The spreadsheet is wip. Please feel free to add more comments. @Alejandro Metke Thanks for the input. If there's an URI for HPO available, which is already being used, I think we should go with this one
Andrew Patterson (Oct 17 2018 at 08:04):
Ok. Thanks. The context is Alejandro and I (who work together) went looking for a suitable HPO codesystem URI a few months back - and I looked through various docs but only found as you say, OIDS. When I saw the one URI assigned in the spreadsheet yesterday - I just didn't know if that had been assigned a long time ago or was a wip.
I guess our argument is that inherent to OWL based ontologies is a xml:base URI that the OWL file itself defines - which seems to be a suitable candidate for code system URIS.
Andrew Patterson (Oct 17 2018 at 08:07):
So we do have a HPO CodeSystem loaded into the CSIRO Ontoserver terminlogy service - using the OWL URI - but that is not being used in any production sense currently. So happy to change - it is not like it is out in the wild being used in lots of places.
But I guess we are putting the argument forward that OWL ontologies should have CodeSystem = their base URI - unless there are good arguments out there as to why that wouldn't work in general..
Patrick Werner (Oct 17 2018 at 14:46):
If we define the CodeSystem URI in the FHIR namespace we can provide some guidance, the actual CodeSystem and derived ValueSets under this URL.
Marc de Graauw (Nov 16 2018 at 11:26):
I believe this (http://www.obofoundry.org/) is the authoritative source for the URI's HPO and a whole lot of other OWL-based biomedical ontologies use in their OWL representation. Would be good if they'd be incorporated in FHIR using the same mechanism for all. For codes, they use both underscore (as in: 'rdf:about="http://purl.obolibrary.org/obo/HP_0000356') and colons (oboInOwl:id is HP:0000356). Not sure why.
Jim Steel (Nov 16 2018 at 13:01):
@Alejandro Metke
Last updated: Apr 12 2022 at 19:14 UTC
