Stream: genomics
Topic: Variation-code use
Hayden Bader (Mar 05 2021 at 20:39):
If we have a variant that we have associated with both a DB ID and an HGVS nomenclature, would it be ok to document it this way?
component(variation-code)
|--code
--|--valueCodeableConcept
------|--coding
------|--system= DB system (i.e. Clinvar)
------|--code = DB code (i.e. Clinvar id)
--|--text = HGVS data
If not, where should that type of information live?
Edit: Better format the component
Kevin Power (Mar 08 2021 at 01:44):
I suppose you could put the HGVS in text but not sure why you would ? We have other component slices for HGVS expressions.
Rachel Kutner (Mar 08 2021 at 13:57):
Which slices?
There is a DNA Change which (if it corresponds to the HL7V2) corresponds specifically to transcript change. AA Change corresponds to Amino Acid Change, and Genomic DNA Change is self evident. But where would an HGVS Name that is the compilation of more than one of these go? Often HGVS names are sent with c. and p. together, for example, and we need a place to send that information.
Kevin Power (Mar 08 2021 at 14:08):
Yea, those are the other slices I am talking about. I am not sure we had considered when a single HGVS representation includes a combo (like c. and p. together). I might have to go remind myself, but is that valid HGVS? Or is it a short-cut for commonly known variants? Where have you put the combo names in V2?
Jamie Jones (Mar 08 2021 at 14:16):
Using hgvs as "display" in the codeableConcept has been done before for clinvar IDs that also show an hgvs, and I think it is preferable to "text"
Jamie Jones (Mar 08 2021 at 14:18):
I think we want to push for our hgvs fields to be fully valid hgvs strings but that display spot could be used for whatever is most convenient (assuming a canonical display is not provided by the code system)
Kevin Power (Mar 08 2021 at 14:30):
Agree with Jamie. The example found here: https://www.ncbi.nlm.nih.gov/clinvar/variation/53158/
Would look something like this:
"valueCodeableConcept" : {
"coding" : [
{
"system" : "http://www.ncbi.nlm.nih.gov/clinvar",
"code" : "53158",
"display" : "NM_000492.3(CFTR):c.-9_14del (p.Met1fs)"
}
]
}
Hayden Bader (Mar 08 2021 at 14:33):
All right. That makes sense. Do we have any recommendation for what should go into the text elements of the various component codeable concepts in the Observation Variant profile?
Kevin Power (Mar 08 2021 at 14:54):
We haven't documented guidance for when/if to use text, so I guess we default to the FHIR guidance.
Rachel Kutner (Mar 15 2021 at 13:42):
HGVS often will be sent like this: RefSeq.X(Gene):c.123X>Y (p.X213Y)
Where each piece is a valid HGVS term (though protein sequence may be missing the refseq), and they string them together. This is very common, and we need a place to be able to send this information if it's stored without explicit relation to an external database.
Where should we send this information in these cases? We cannot send it here in the display field without a code/code system, can we? In that case, my expectation would have been to send it via Text, as I don't see another place to send it.
Kevin Power (Mar 15 2021 at 17:07):
To me, the first question to answer is this: Is the HGVS expression Rachel references above a valid HGVS expression? Or, is it a HGVS like expression some systems use? I think the answer to that question can help guide us in determining where to put it.
Bret H (Apr 01 2021 at 13:10):
@Rachel Kutner you can always break the HGVS down to it's component pieces and send with the discrete components for position, reference etc..
Rachel Kutner (Apr 01 2021 at 13:42):
We have other values that are sent in those fields. If we send multiple, is there a way to tell which go with which piece when receiving the message, vs which were recorded separately?
Last updated: Apr 12 2022 at 19:14 UTC
