GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-03 > 1269874869


From:
Subject: Re: [DNA] 23andme cousin results
Date: Mon, 29 Mar 2010 11:06:11 EDT


23andMe is able to apply a correction factor for Ashkenazi Jews (which
still may not be quite enough, but is a start). This is based on running their
simulations using a gene pool of people who identified all four grandparents
as Ashkenazi on the "Where Do You Come From?" survey. The simulation showed
the percentage of cases where a segment came out of the general gene pool
rather than the simulated pedigrees. I doubt that FTDNA has the wherewithal to
do a similar thing.

For the time being, other endogamous groups will need to study their own
results, but if everyone will answer the survey, that might ultimately provide
a useful databank.

Ann Turner

In a message dated 3/29/2010 3:37:31 AM Pacific Daylight Time,
writes:

> I think these 65% to 75% identical measures is based on assumption of
> independent SNP's not block structures, but you are right on the spot that
> parents with ancestry from same etnicity or geographical region likely will
> have collection of blocks that has a common source, so most of your cousins
> or any seemingly totally unrelated random individual with same or related
> ethicity from a overlapping geographical region. This messes up 23andme's
> cousin calculations that are only correct assuming total unrelatedness but
> this is not a correct assumption within related enticities or geographical
> regions. This result in overestimation of cousin relationships. This means
> some kind of correction factor is needed when comparing.


This thread: