FHIR Chat · Compound heterozygosity

Stream: genomics

Topic: Compound heterozygosity

Bob Dolin (Aug 13 2020 at 22:14):

Hi, hoping to get a sanity check here. I have a VCF file, where the patient has a different variant allele on each homolog (G>GT on one chromosome, and G>GTT on homologous chromosome):

#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample
7 117163435 . G GT,GTT . . . GT 1/2

Would folks generally represent this with two variant profiles (a heterozygous GT and a heterozygous GTT)? Would anyone consider representing this as a single variant, where ALT="GT,GTT" and allelic state is something like 'compound heterozygous'? I've been using the former representation, but just curious...

Kevin Power (Aug 14 2020 at 13:44):

Sounds like some more guidance we are missing :smile:

I would have probably sent this as two variants as well. Suppose you could even add a SequencePhaseRelationship with the two as well, since you can say for sure they are trans?

Jamie Jones (Aug 14 2020 at 13:46):

We have a compound het example that shows how to do it both ways (though uses hgvs style)

Kevin Power (Aug 14 2020 at 13:51):

So not 1 or 2 variants, but 3 @Bob Dolin :smile:

Kevin Power (Aug 14 2020 at 13:52):

Glad @Jamie Jones reminded about that option though, I had already forgot about the Complex Variant Type component. Dang.

Jamie Jones (Aug 14 2020 at 13:53):

Well aren't we considering pulling complex variant type? I agree the list needs work but compound het needs a home

Jamie Jones (Aug 14 2020 at 13:53):

Single variant guidance may be cleanest...

Bob Dolin (Aug 14 2020 at 14:13):

@Kevin Power @Jamie Jones Thanks for your comments.

Kevin Power (Aug 14 2020 at 14:14):

Please listen to @Jamie Jones more than me :smile:

Joel Schneider (Dec 21 2020 at 17:06):

My $0.02 is, if we don't currently have a solid clinical use case for compound heterozygosity, maybe add a disclaimer saying we haven't fully addressed it and welcome comments.

Rachel Kutner (Dec 22 2020 at 15:18):

My two cents:

Variants that are suspected/potential Compound Heteozygous make most sense reported separately.
A lab stated that they would show the variants together on the same line once the phase has been established as trans, but the way they display the variants looks like they are separate variants being assessed together, rather than a single variant (since they often aren't at the exact same position).

If they need to be "connected" somehow, it seems that the Interpretation/Annotation is where to connect them, via the Phase and the shared Associated Phenotype (like has already been mentioned above).

Each Variant represents "what was found". This can include variant-specific information that doesn't take into account other context (including other variants found on the patient)
The annotation/interpretation layer over that answers "how does this result affect the patient".

I've thought about this a fair bit because there are 2 layers we need to clearly represent so as not to create confusion. The population-level assessment of a variant (pathogenic, benign, etc) vs the patient-level assessment (is this patient affected by the variant - this might need to take into account more than just genomic information, more than one variant result, etc.). I haven't yet found a great differentiation for these types of information, so my suggestion comes from my brainstorming what might be a nice solution for disambiguating that information.