FHIR Chat · Did you find this variant vs What variant did you find?

Stream: genomics

Topic: Did you find this variant vs What variant did you find?

Bob Milius (Oct 19 2018 at 17:29):

In the Complex Variant Observation in our current build of the IG, the observation.code is bound to "Genetic variant assessment" (69548-6) with an answer list of "Present|Absent|No call|Indeterminate" (LL1971-2). So this is asking the question, "Is this complex variant present?" I need to ask the question "What is the complex variant present?" I can use this profile by always using "Present" in the observation.valueCodeableConcept, and putting in the HLA allele name as a component for "Complex genetic variant code" (81261-0). It's definitely doable and I'll use that if that's what is needed to be conformant to the IG, but that seems a bit contrived (that really isn't the question I asked). Is there any problem with me making an observation with observation.code being the "Complex genetic variant code" with the value being the HLA allele? It wouldn't adhere to any of the profiles in the IG, but it would be a valid FHIR Observation.

So right now, with the current Complex Variant Observation, I'm using this...

...
<code>
    <coding>
        <system value="http://loinc.org"/>
        <code value="69548-6"/>
        <display value="Genetic variant assessment"/>
    </coding>
</code>
...
<valueCodeableConcept>
    <coding>
        <system value="http://loinc.org"/>
        <code value="LA9633-4"/>
        <display value="Present"/> <!-- always Present for this question -->
    </coding>
</valueCodeableConcept>
...
<component>
    <code>
        <coding>
            <system value="http://loinc.org"/>
            <code value="48018-6"/>
            <display value="Gene Studied"/>
        </coding>
    </code>
    <valueCodeableConcept>
        <coding>
            <system value="https://www.genenames.org/"/>
            <code value="HGNC:4931"/>
            <display value="HLA-A"/>
        </coding>
        <text value="HLA-A"/>
    </valueCodeableConcept>
</component>
<component>
    <code>
        <coding>
            <system value="http://loinc.org"/>
            <code value="81261-0"/>
            <display value="Complex variant name Nominal"/>
        </coding>
        <coding>
            <system value="http://loinc.org"/>
            <code value="57290-9"/>
            <display value="HLA-A [Type] by High resolution"/>
        </coding>
    </code>
    <valueCodeableConcept>
        <coding>
            <system value="http://www.ebi.ac.uk/ipd/imgt/hla/"/>
            <version value="3.31.0"/>
            <code value="HLA-A*03:01:01:01"/>
            <display value="HLA-A*03:01:01:01"/>
        </coding>
    </valueCodeableConcept>
</component>

Here's an example of the code/value/components the new observation would have. It's shorter and seems cleaner:

<code>
    <coding>
        <system value="http://loinc.org"/>
        <code value="81261-0"/>
        <display value="Complex variant name Nominal"/>
    </coding>
    <coding>
        <system value="http://loinc.org"/>
        <code value="57290-9"/>
        <display value="HLA-A [Type] by High resolution"/>
    </coding>
</code>
...
<valueCodeableConcept>
    <coding>
        <system value="http://www.ebi.ac.uk/ipd/imgt/hla/"/>
        <version value="3.31.0"/>
        <code value="HLA-A*03:01:01:01"/>
        <display value="HLA-A*03:01:01:01"/>
    </coding>
</valueCodeableConcept>
...
<component>
    <code>
        <coding>
            <system value="http://loinc.org"/>
            <code value="48018-6"/>
            <display value="Gene Studied"/>
        </coding>
    </code>
    <valueCodeableConcept>
        <coding>
            <system value="https://www.genenames.org/"/>
            <code value="HGNC:4931"/>
            <display value="HLA-A"/>
        </coding>
    </valueCodeableConcept>
</component>

Kevin Power (Oct 19 2018 at 18:04):

DescribedVariant is the same way, and it allows you to say "I looked for a variant but did not find it" - which is important in some contexts. Also, not all variants will be described by a single code as nicely as you have in the HLA Allele world.

Not to say we have to always do it that way, but that is why it is that way today.

Bob Milius (Oct 19 2018 at 18:32):

Yes, I'm not saying "Do you have this variant?" isn't important, especially for PGx. For HLA, one often asks the question, "Is HLA-B*15:02 present?" which is important to know if you need to determine carbamazepine hypersensitivity. But that's a different question than, "What HLA-B variant(s) is/are present?" Both are important questions. But they are different questions.

Bob Milius (Oct 19 2018 at 18:40):

The first is important to inform potential drug sensitivity, the second is for tissue matching

Kevin Power (Oct 19 2018 at 18:44):

We could propose a different profile or two to reflect how the question was being asked? I don't want to proliferate profiles, but maybe it is important here?

Bob Milius (Oct 19 2018 at 19:04):

I can work with what we have, but it would nice to be able to articulate the real question. I've thought of making an HLA specific observation profile for this, but would rather re-use the IG as much as possible. Maybe this is related to the "What was found?" discussion?

Bob Milius (Oct 19 2018 at 19:12):

or maybe go back to Observation for Haplotype, and add Haplotype to the list of possible derivedFrom (i.e., Reference(Cytogenic Notation | Complex variation | Described Variant | Haplotype). Not perfect, but then I could use Haplotype the same way as @Bob Dolin uses it for PGx, and still use it for haplotypes of alleles for a set of HLA genes (eg HLA-A01:01:01:01~HLA-B15:02:02)

Bob Milius (Oct 19 2018 at 19:12):

Haplotype of Haplotypes

Kevin Power (Oct 19 2018 at 19:13):

I thought at Baltimore we decided that you should just create some HLA specific profiles for your use cases? Are you reconsidering that?

Bob Dolin (Oct 19 2018 at 19:13):

I've been thinking it might be good, once the IG is complete, for us to define a set of genomics FHIR Operations. That might help here, by letting the questions vary as needed, while still returning responses based on the existing profiles?

Bob Milius (Oct 19 2018 at 19:17):

@Kevin Power That's what I've working on. Mostly constraining the current IG to our needs. This one came up where I'm considering another observation profile for HLA for this question. My hesitation is that the Complex Variant ID is one of the components in the Complex Variant Profile. I'm weary about sending the same info (sorta) in a different structure. Nice to have consistency.

Bob Milius (Oct 19 2018 at 19:19):

For now I think I'll use the Complex Variant profile as it is. I'll also add a ticket to change add Haplotype to the derivedFrom list in Haplotype.

Kevin Power (Oct 19 2018 at 19:21):

OK - what other constraints have you come up with thus far?

Bob Milius (Oct 19 2018 at 19:30):

mostly adding codings for specific HLA typings (eg 57290-9, "HLA-A [Type] by High resolution"), requiring use of imgt/hla and version for describing single HLA alleles, and glstring.org for haplotypes, genotypes, and ambiguous allele lists; describing how sequence resource is used with allele sequence database, etc. If using the complex variant profile, require adding complex variant name and gene name components. Mostly requiring specific bindings and additional codings for same concepts that are already required.

Bob Milius (Oct 19 2018 at 19:30):

stuff like that

Kevin Power (Oct 19 2018 at 19:32):

Makes sense. Are you creating profiles to document those constraints?

Bob Milius (Oct 19 2018 at 19:36):

I'm working on that now. I want to update the HLA section of the CG IG using this work to inform (won't have HLA profiles in the CG IG, but just a discussion of how this can be captured. This is why I'm trying to reuse the CG IG as much as possible. Also working on examples of resources that are separate from the bundle I had originally added to the IG.

Kevin Power (Oct 19 2018 at 19:39):

Sounds great. And speaking of examples, there are some pending tracker items to update the HLA examples in the IG. Can you take those on if I send them your way? Or, do we possibly remove it from the current IG while you are working?