Does CNVs are genetically inherited?

Question

[peris]

I have called CNVs from exomes, in multiple members of distant cousins suffering from same disease using Conifer in different families. Then I tried to identify the region which are inherited by all the cousins. But I did not find any regions which are shared among themselves. But, if I try to look for the shared CNVs withing all the 50 samples, there exist many shared regions, the samples being from different families. Can I make any interpretation from this observation? Does the CNVs follow Mendelian inheritance?

[Cliff Beall]

If a CNV is due to a duplication it might involve multiple chromosomal locations and might have non-Mendelian inheritance. Deletions and tandem duplications should be Mendelian.

Not sure if this is the answer you need - it seems like you might not be asking about Mendelian inheritance but just genetic inheritance in general.

[peris]

Thanks for the comment. Its clear one of my doubt.

Answer 1

[Kizuna]

at least in Mendelian diseases, CNVs follow Mendel Laws.

If your distant cousins suffer from the same disease, and if this disease is mendelian (most probably your case), you should have same CNVs.. de novo CNVs can arise, but this is really rare.

To the best of my knowledge, a CNV shared between 50 subjects (even if they are all patients), is not disease causing... only if you have a founder effect.. because CNVs are really rare...

P.S: a CoNIFER analysis alone is not sufficient to detect a CNV.. a Sanger sequencing step is always the gold standard for validating a " candidate CNV"

I would advise you to do a co-segregation analyses in trio at least (Father, mother and affected sibling)..

Hope this would help.

Kiz

[DG]

Just to quibble but I don't think we should say that CNVs are rare, it really appears that CNVs are quite common, and just like other forms of mutations many of them are neutral. The number causing a mendelian disease is small compared to small indels and SNVs but that is because until recently CNVs were much harder to test and look for except the larger abnormalities. The current stats are a function of our ability to detect and classify CNVs, and do not accurately reflect biological reality.

There are also plenty of common CNVs observed in populations, just the same as SNVs.

[Kizuna]

I do agree,

however, I should point out to the fact that common CNVs are called CNPs to the best of my knowledge (I might be wrong), and surely they do not cause mendelian diseases (Peris case)..

CNVs in mendelian diseases are known to be prevalent in like 1% of the cases :)

[DG]

Absolutely common CNV's shouldn't be causing Mendelian disease, and I wasn't implying that they do. As for nomenclature CNV (like SNV) should be the broadest possible term and encompasess both polymorphisms and pathogenic variants.

My point actually more follows on the point about CNVs being implicated in about 1% of cases. I think in the literature it is actually probably a higher estimate at the moment (since 80% of known Mendelian diseases are caused by "simple" SNVs and small indels) but that aside, it is still a function of most Mendelian disease genes discovered currently being the "low-hanging fruit." Structural variation still remains a more difficult challenge to identify within the constraints of how most mendelian disease investigations are currently being done. Mostly exomes and mostly looking for SNVs and indels.

[Alex Paciorkowski]

A couple of inaccurate things in your post @Kizuna:

1. A CNV shared between 50 subjects (with the same disease) could very well be disease causing. This happens all the time. Deletion 22q11 syndrome. Williams syndrome. The list is very long of examples in human genetic disease.
2. I'm not sure what your definitions of "really rare" vs "common" are. CNVs in the human genome in general are really, really common -- this is one of the interesting discoveries about the human genome in the past 10 years. Furthermore, de novo CNVs can be common causes of certain diseases -- i.e. up to 10% of autism is caused by pathogenic copy number variants. Most geneticists would agree that "rare" is something less than 1 in 100,000. CNVs in general are far more common than that.
3. I don't know why you recommend Sanger sequencing to detect copy number variants. Sanger is notoriously unable to detect deleted alleles. Most would recommend quantitative PCR, not Sanger methods.
4. Common CNVs (for example those in the Database of Genomic Variants) are called... common CNVs. I've never heard of the acronym "CNP." I would also point out that several CNVs are well described in the literature to be subject to variable penetrance, and pathogenicity (or lack of pathogenicity) is not as clear-cut as you suggest.

[peris]

Thanks Kiz for the answer. Its really helpful.

Answer 2

[Nicholas Blackburn]

I've not used Conifer so I don't know how it identifies and outputs CNVs.

Could it be that the CNV chromosome positions differ slightly between each person in the family? So when you're looking for the inherited region, it's there but just slightly different position callings in each cousin?

I'd make BED files for each cousin and then use bedtools to intersect them Bedtools Compare Multiple Bed Files? to find the overlapping CNV regions.

[peris]

Thanks for the advice. I will follow it.