GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-03 > 1269405003


From: David Faux <>
Subject: Re: [DNA] Comparing Ancestry Painting (23andme) to AncestralOrigins Genome Browser (deCODEme)
Date: Tue, 23 Mar 2010 21:31:25 -0700
References: <da404fe11003232038l44839595m6755fa76e0bcfbd0@mail.gmail.com>
In-Reply-To: <da404fe11003232038l44839595m6755fa76e0bcfbd0@mail.gmail.com>


Steven,

Unfortuantely comparisons between 23andMe and deCODEme often appear to be in
the apples to oranges category. Many individuals obtain 100% European on
23andMe, which I maintain is due to the archaic algorithm being used as well
as only three reference groups (whereas Dr. McDonald uses all 52 CEPH-HDGP
populations in his analysis). Hence a fair number of false negatives but
very few false positives (my opinion only but based on a lot of samples with
known genealogies).

deCODEme tends to inflate Asian and African for those who are primarily or
entirely European. Hence there are many Europeans with over 10% Asian on
the X, but no blocks that cross the midline or in the case of males cross
from side to side.

As an example of the inconsistency and the fact that most will get different
biogeographic results with deCODEme and 23andMe I will note my cousin's
results. 23andMe detected a small amount of African in his uncle but none
in him. However when my cousin's data was uploaded to deCODEme there
appeared a very large (15 Mb) African block which was actually a rectangle
that reached the mid point of the single chromosome 12 (meaning a clear
African block on one of his two chromsomes). Clearly there is something
there and it stood out even more dramatically in the browser. I have no
idea why 23andMe did not detect this distinctive African segment. I tried
Firefox and the Firefly program to "look into" the 23andMe Ancestry Painting
and see the coding. There was no African at all noted on any chromosome.

I could give other examples, but I honestly don't know which company output
to trust.

David K. Faux.
On Tue, Mar 23, 2010 at 8:38 PM, steven perkins <> wrote:

> Hello:
>
> In 23andme my Ancestry Painting is 100% European and my 1st cousin is
> 99%European and 1% African. At deCODEme in Ancestral Origins I am
> 2%African, 4% East Asian, and 94% European on Chromosomes 1-22, and
> 11% African, 1% East Asian and 88% European on the X Chromosome. I
> don't have the readings from my cousin yet, but I compared our
> ancestry using the Genome Browser and I am noticeably more African
> than she is on all but one chromosome, and much more on the X. She has
> much more East Asian results on all but one chromosome.
>
> I am wondering how 23andme finds African in her Ancestry Painting but
> not in mine. Anyone know how the systems differ?
>
> Does anyone have results from all three autosomal tests: 23andme,
> decodeme, and FTDNA? Any differences in biogeographic ancestry
> between the tests?
>
> Thanks,
>
> Steven
>
>


This thread: