Stream: genomics
Topic: Region Studied Implementation
Jamie Jones (Jan 13 2019 at 17:03):
Hey all, looking at using the recently added http://build.fhir.org/ig/HL7/genomics-reporting/obs-region-studied.html for querying on specific genes during the connectathon, some questions came up around the suggested valueset for the code and use of the region-scope component vs Observation.method. Please feel free to elaborate @Patrick Werner
Bret H (Jan 13 2019 at 17:24):
Here is a proposed example meeting the use case of providing the observations for a specific gene and specific patient:
{
"resourceType": "Observation",
"id": "studReg",
"meta": {
"profile": [
"http://hl7.org/fhir/uv/genomics-reporting/StructureDefinition/obs-region-studied"
]
},
"status": "final",
"category": {
"coding": [
{"coding":[{"system":"http://loinc.org","code":"TBD-RegionsStudied"}]}
]
},
"code": {
"coding": [
{
"system": "http://loinc.org",
"code": "TBD-Described",
"display": "Genetic variant assessment"
}
]
},
"subject": {
"reference": "Patient/HG03239"
},
"issued": "2019-01-07T17:48:34+05:30",
"valueCodeableConcept": {
"coding": [
{
"system": "willBeRemoved",
"code": "None",
"display": "NotNeeded"
}
]
},
"specimen": {
"reference": "Specimen/specimen1"
},
"component": [
{
"code": {
"coding": [
{"coding":[{"system":"http://loinc.org","code":"48018-6"}]}
]
},
"valueCodeableConcept": {
"coding": [
{
"system": "http://www.genenames.org",
"code": "HGNC:2625",
"display": "CYP2D6"
}
]
}
},
{
"code": {
"coding": [
{"coding":[{"system":"http://loinc.org","code":"TBD-RegionScope"}]}
]
},
"valueCodeableConcept": {
"coding": [
{
"system:"CG"
"code": "SpecificVariants"
}
]
}
}
]
"hasMember":[
{
"reference":"Observation/descrete1"
},
{
"reference":"Observation/descrete2"
}
]
}
Jamie Jones (Jan 13 2019 at 17:27):
corresponding query draft: Observation?code=TBD-RegionsStudied&combo-code=http://loinc.org|48018-6&combo-value-concept=http://www.genenames.org|HGNC:2625
Kevin Power (Jan 13 2019 at 17:39):
So, you are asking about what would be the RegionStudied.method versus RegionStudied.component[region scope].valueCodeableConcept? @James Jones
Patrick Werner (Jan 13 2019 at 18:14):
So, you are asking about what would be the RegionStudied.method versus RegionStudied.component[region scope].valueCodeableConcept? @James Jones
yes. That was the question. I would be in favor to use method and get rid of the component as it duplicates method and could lead to redundant data.
Kevin Power (Jan 13 2019 at 18:16):
Not sure I agree, but am OK with a tracker to discuss. The "scope" was meant to be a simpler way to understand the region, as methods will vary quite a bit (I think?).
Patrick Werner (Jan 13 2019 at 18:29):
I'll create a tracker now.
Kevin Power (Jan 13 2019 at 18:36):
Can you give me some examples of what we will use for method? @Bret H
Patrick Werner (Jan 13 2019 at 18:41):
created GF#19975
but i'm now also unsure if it is the same. Depends on what will be included in method.
Bret H (Jan 13 2019 at 18:52):
@Kevin Power already present are 'Exome, Genome, SpecificVariant...' these are not really methods but conceptual context of what was done. E.g. 'PCR' is a method but could be whole-gene or specific variant. You'd need the PCR probes to infer OR be told. FYI: @James Jones suggested context when I was looking for a word.
Scope is not quite the right word and it is not a broad description of the method but related to the methods used to study, in my opinion.
Bret H (Jan 13 2019 at 18:53):
pasted image
We need to add these to the example. @Bob Dolin sent me the specific described variant profiles that would be used for the 'has member' relationship
Kevin Power (Jan 13 2019 at 18:57):
So you are saying we need something like “scope” but the name isn’t right? Your example of PCR is why I was just thinking we needed “scope”. I like the name “context” as well if that makes more sense ?
Bret H (Jan 13 2019 at 19:03):
be interested in Patrick's (and any others comments)...but yeah, 'scope' to me is not quite right. But this could be word smithing as 'context' kinda says the same as 'scope'
Jamie Jones (Jan 13 2019 at 19:08):
I think the question here is if it's meant as a summary/clarification of the method or trying to say something of its own, perhaps from where the regionstudied observation was created? (we are looking at building these observations from a query, perhaps we should be using GDR for this though...)
Bob Dolin (Jan 13 2019 at 20:18):
main issue I have with Region-studied profile (ballot comment #19845) is that we need to have the ability to assert arbitrary locations. Main use cases that come to mind are: [1] region studied may be intergenic; [2] differentiating variant calling accuracy across sub-regions of a gene.
Bob Dolin (Jan 13 2019 at 20:20):
A bit more explanation on [2] - I can't simply say that gene TPMT was studied. It might be that I only studied the exons. It might be that for technical reasons, portions of the gene couldn't be assessed. But I need that level of granularity in order to assess the accuracy of variant calls, determine where there might be false-negatives, etc.
Kevin Power (Jan 13 2019 at 20:32):
@Bob Dolin - those use cases make sense - we just can't do it yet. We can discuss those when we get to that ballot comment. If you have a proposal, that would help the discussion about that ballot comment significantly.
Bob Dolin (Jan 13 2019 at 21:08):
@Kevin Power Thanks Kevin. I stopped short of including a proposal in the ballot comment, other than to suggest we look at GA4GH's htsget protocol, which supports location-based queries. From there, I feel like we have to think through the design a bit. Location start and end position alone isn't sufficient - it needs to be coupled with chromosome and build, or with reference sequence, along with coordinate system.
Last updated: Apr 12 2022 at 19:14 UTC
