GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2009-12 > 1260231480
Subject: Re: [DNA] real relatedness or fluff and fairytales?
Date: Mon, 7 Dec 2009 19:18:00 EST
The centromeres are SNP "deserts", but 23andMe requires a certain threshold
of cM and SNPs before identifying a haploblock. Currently at least one
haploblock must contain 700 SNPs and 7 cM. That eliminates most of the artifacts
around the centromeres.
One "wrong SNP" call (i.e. opposite homozygotes) is allowed if it is
surrounded by hundreds of SNPs in a long continuous run.
For a calibration of error rates, I looked for Mendelian inconsistencies in
two parent-child trios from my own family study.These would be SNPs where
the parent is say "AA" and the child is "GG". Here is the tally (out of
Mother and son: 13
Father and son: 34 (6 in a cluster)
Mother and son: 17
Father and son: 25 (5 in a cluster)
The SNPs that show up in a cluster could signify a microdeletion, where one
chromosome is missing some bases. If 23andMe sees just one allele, it gets
called as homozygous.
For 23andMe customers with parent-child data, you can look at the table
view in Ancestry Labs to see if any chromosome has more than one segment. If
the microdeletion is big enough, it will also be visible as a small white area
in the Family Inheritance view.
In a message dated 12/6/2009 11:17:33 AM Pacific Standard Time,
> Note that 23andMe also ignores the centromere regions completely.
> This makes the calculations easy for them but may introduce some severe
> length misinterpretations when a haploblock stretches over the
> centromeric region.
> Also there is apparently a threshold for "wrong SNP" calls. If you
> compare a parent-child result closely you will find a couple of SNPs
> that don't share any of the alleles. 23andMe simply considers them as
> typing errors or mutations and ignores them. This is in my opinion very
> dangerous (especially in highly conserved regions). The correct way to
> handle this would be to at least list the inconsistencies and if
> possible give a reason why they are ignored. So for now I recommend to
> not simply trust the display of a continuous haploblock in the 23andMe
> chromosome browser. You should definitely verify the raw SNP calls.
> Another phenomenon that cannot be understood from a simple SNP chip
> result are large scale re-arrangements of large chromosome stretches. We
> know already from the Y chromosome that we have a significant number of
> people who have duplications, inversions or deletions in palindromic
> regions of the chromosome. This of course affects the length and the
> pattern of haploblocks and this will definitely affect the "half
> identical regions".