Stream: genomics
Topic: Corrections of Terminology URIs in Core
Patrick Werner (Nov 04 2018 at 11:51):
I've just made a GF item and PR to correct the identified errors in the existing Terminology URIs:
https://chat.fhir.org/#narrow/stream/48-terminology/subject/corrections.20for.20CG.20terminologsy.20URIs/near/205077
Patrick Werner (Nov 04 2018 at 11:52):
@Julian Sass and myself will continue the CodeSystem work. Bringing the corrections of existing URIs into R4 was the first step/priority.
Kevin Power (Nov 04 2018 at 16:19):
Looks good to me
Patrick Werner (Nov 05 2018 at 18:54):
Unfortunately the tracker was set to deferredby Melva today :anguished:
Lloyd McKenzie (Nov 05 2018 at 19:04):
All new trackers are automatically deferred. Work groups are still free to take them up if they wish.
Lloyd McKenzie (Nov 05 2018 at 19:05):
You can badger the Vocab co-chairs to still take it up. (Rob Hausam is probably the most appropriate target)
Kevin Power (Nov 05 2018 at 19:06):
I think Patrick @'ed Rob on the pull request.
Patrick Werner (Nov 05 2018 at 19:53):
Yes i did. Thanks for the explanation
Patrick Werner (Nov 07 2018 at 17:10):
Here is Robs response to the PR:
I'm wondering if the description for 'refseq' should be "RefSeq: NCBI Reference Sequence Database", as it is shown on the NLM website, rather than "REFSEQ : National Center for Biotechnology Information (NCBI) Reference Sequences" as you currently have it? Maybe that doesn't matter, particularly if you have a different source for the description that you used. Regarding 'http' vs. 'https' for refseq, I think it's not entirely clear which one you should use here. I note that 'http://www.ncbi.nlm.nih.gov/refseq/ is automatically redirected to 'https://www.ncbi.nlm.nih.gov/refseq/', so I'm not sure that specifying it with 'http' actually helps or is the best choice. All of the other terminologies in terminologies-systems currently are 'http', but I don't think there is an absolute rule about that (at least not that I'm thinking of), and the general rule is that if the publisher defines a url then we use that one. Another thing that I see is that http://www.ncbi.nlm.nih.gov/clinvar/varId currently doesn't resolve, whereas http://www.ncbi.nlm.nih.gov/clinvar does. Is that what you expect? In general it's good practice, of course, to use urls that do actually resolve to something useful. I also don't understand why adding "PharmGKB Accession ID" is useful (or even correct?) in the context where you added it?
Patrick Werner (Nov 07 2018 at 17:18):
Regarding the points:
1) Naming of RefSeq: To me both terms are equivalent, but i agree with Rob to alter the description to the version of the NCBI Homepage
2) to align with other CodeSystem SIDs i would have the refseq URI starting with http. It still resolves (with a redirect)
3) http://www.ncbi.nlm.nih.gov/clinvar/varId doesnt resolve. But we need to distinguish between clinvar varIds and allele IDs as these both have integer codes which are overlapping. An alternative would be to define our own URI (e.g. http://hl7.org/fhir/sid/clinvar/varid). To keep the updated URI closer to the former clinvar SID i would prefer adding the path varid to the existing URI
4) We added "PharmGKB Accession ID" to PharmGKB to express that this Codesystem contains the Accession ID and not every single term (which are not coded) of PharmGKB
@Julian Sass @Kevin Power @Rob Hausam
Kevin Power (Nov 07 2018 at 17:33):
1) Agree with Rob
2) Whichever aligns best with how others are presented on the page.
3) Could we put "Variant ID" or "Var ID" in the comment column to help make the distinction more simply for now, since that is sort of why we added the comment mentioned in #4.
4) In lieu of a more specific URI, it seems that adding the comment helps clarify?
Patrick Werner (Nov 07 2018 at 17:38):
@3) The problem will then re-arise when we define a CodeSystem for the allel-IDs
Patrick Werner (Nov 07 2018 at 17:39):
If we make the change now we then could simply use http://www.ncbi.nlm.nih.gov/clinvar/varId and http://www.ncbi.nlm.nih.gov/clinvar/allelid
Patrick Werner (Nov 07 2018 at 17:40):
if we decide, that our URIs should always resolve we could use http://hl7.org/fhir/sid/clinvar/varid and http://hl7.org/fhir/sid/clinvar/allelid
Kevin Power (Nov 07 2018 at 17:45):
I have heard alleleid is going away (someday). I prefer to have a resolvable URI, but I don't have strong feelings.
Kevin Power (Nov 07 2018 at 17:50):
Can we make a reference to a page like this to help show the differences? https://www.ncbi.nlm.nih.gov/clinvar/docs/variation_report/#allele
Rob Hausam (Nov 07 2018 at 18:32):
I'll just add that we shouldn't be using an NLM url unless NLM is aware of and agrees to that (whether the url itself actually resolves to something useful at this point, or not). I'm guessing that NLM hasn't been involved in discussion so far about this (but I could be wrong on that)?
Kevin Power (Nov 07 2018 at 18:43):
The current URL has been included for quite some time, but I am not sure if we talked to them about it or not. Perhaps @Gil Alterovitz can answer that? CC: @James Jones
Jamie Jones (Nov 07 2018 at 19:10):
I believe Clem and Liz could be brought in on this pretty easily if there are any issues with it. They were both at the WGM when we discussed some of these urls as well.
Kevin Power (Nov 07 2018 at 19:35):
@Rob Hausam - what sort of agreement do we need? Email from Clem? Written permission?
Rob Hausam (Nov 07 2018 at 20:32):
I think any kind of acknowledgement from them that they are aware and agree with using these urls and that at least has some notion of being from an official source should be sufficient. So an email from Clem or whoever might be in charge of such things I think would be fine.
Kevin Power (Nov 07 2018 at 20:54):
Email sent to Clem and Liz (and CC'ed Rob, Patrick, Jamie)
Kevin Power (Nov 07 2018 at 20:59):
Regarding a URI for ClinVar, I think I lean towards this approach: http://hl7.org/fhir/sid/clinvar/varid to set up for 'alleleid' later if we need it. We will still have the source with the real link.
Kevin Power (Nov 07 2018 at 21:09):
And perhaps we should even do something like http://hl7.org/fhir/side/pharmgkb/accessionid to be more clear on the PharmGKB row?
Last updated: Apr 12 2022 at 19:14 UTC
