CompHet question

[lindakjcao]

Hi Brentp,

I have another question regarding the CompHet tagging. Basically, what you have done in the script has no problem. But in our lab (we are clinical diagnostic lab), we have a bit more relaxed definition of CompHet that even the variant in one parent is Homo change, we would still count the two het variants in the same gene in the kid as CompHet, especially when the gene is in ACMG secondary finding list. Here is one example, the gene HSPG2 is involved in the phenotypes of developmental delayed. And we found:

Using Slivar, the variants in proband were not be tagged as CompHet, but in our case, we consider the variants CompHet because of the gene.
Is there any way we can still utilize your CompHet function to flag the untypical CompHet variants like this? Thank you.

Linda

[brentp]

Hi Linda, that's currently not allowed. The compound-hets tool assumes a trio (or duo with flag) with unaffected parents. For your case, is the father affected? Or how do you explain that candidate? I'd certainly like to support use-cases (but always have to consider the complexity of the implementation and the complexity of the user-interface).

[conlinl]

Hi Brent, I work with Linda, and there are several scenarios where we are interested in determining compound heterozygous state even if a parent is homozygous. One concern are variants that are known to be hypomorphs, in that they do not cause serious disease in the homozygous state, but can cause disease when in trans with a pathogenic variant. Another example is that the variants are known to increase risk for disease in the hom state, and are quite common in the general population, so it would not be unusual for an unaffected parent to also be homozygous. We would also be concerned for the scenario of an affected parent being homozygous, while their affect child is compound heterozygous - this is especially in the case of genes with high frequency founder variants and/or consanguineous populations. Here are some clinically important variants that fit each of these scenarios:

1. Hypomorphic allele: BTD p.Asp424His; this variant does not cause disease in the hom state (and it is very common), but does cause disease when in trans with a pathogenic variant. This gene is on the ACMG secondary finding list, and is to be reported when "two hits" exist in this gene. It is very possible for an unaffected homozygous parent to have an affected child with compound heterozygosity. https://preview.ncbi.nlm.nih.gov/clinvar/variation/1900/
2. Common variant with low clinical risk for disease: HFE p.Cys282Tyr associated with high risk for Hemochromatosis; This disease/variant is also on the ACMG secondary finding list to report. The carrier frequency for this variant is ~1%-5% in various populations, so it is not unthinkable that a homozygous parent could have a compound heterozygous child. Penetrance for clinical presentations for this disease is very low (though altered biochemical tests would be detected). https://preview.ncbi.nlm.nih.gov/clinvar/variation/9/ https://www.ncbi.nlm.nih.gov/books/NBK1440/#hemochromatosis.Molecular_Genetics
3. Founder variants in GJB2 and affected parent(s). There are several founder variants in this gene associated with autosomal recessive hearing loss (e.g. c.35del in 2-4% NFE, p.V37I in 10% of Han Chinese/SEA, c.167delT in ~8% of AJ) It would not be unheard of to have an affected parent with homozygosity for one of these founder variants to have an child with compound heterozygosity with a different founder variant, especially if the other parent has a different genetic ancestry. https://www.ncbi.nlm.nih.gov/books/NBK1272/#dfnb1.Molecular_Genetics

I can come up with many other variants, but these were the ones that we see all the time. Please let me know if additional examples would be helpful!

[brentp]

Thank you for the thorough explanation. This helps me to prioritize work. Let me have a look at how to implement this in the next few weeks. It will require substantial changes, so no promises, but you've certainly convinced me of the utility.

[conlinl]

Thanks so much Brent! This is definitely an option that would be helpful for us to use in these contexts! I would assume that others who are working with rare, highly penetrant disorders (with unaffected parents) would be just fine without this cmp het option, and it's these specific scenarios (which depend on your particular family structure and severity/prevalence of the phenotype of interest) where it would be very welcomed!