GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2002-12 > 1040160018


From: "pwreed" <>
Subject: Re: [DNA] AncestrybyDNA results
Date: Tue, 17 Dec 2002 21:20:18 -0000
References: <4.3.2.7.2.20021217110922.00c07500@casbah.acns.nwu.edu>


----- Original Message -----
From: "Joseph B. Lambert"
In other words, are the results additive? If
> so, I could imagine that a person could get a result of 10-15% East
Asian
> just because there was an accumulation of East Asian blood, not
because
> this person had a single, full-blooded great-grandparent. Again, if
so,
> that means that the method is tracking more distant ancestry. I
ask,
> because I have seen the statement that the method is meaningless
beyond
> great-great-grandparents (who each contribute about 6% of our
genome)
> because of the 5% error. If not, I would imagine that significant
(>5%)
> contributions to non-European ancestry can show up by accumulation
from
> very distant, full-blooded ancestors.

You may be referring to the comments I made a few weeks back when I
stated
"the percentages of admixture that can be reliably measured have to be
large (>> 5%) and therefore either very recent or very prevalent in
the general ancestral population."

The "very prevalent in the general ancestral population" statement
covers your example where ALL great-grandparents had say 12% east
Asian. My feeling in that case though would be that there would be at
least a suspicion of non-European ancestry.

Because of the somewhat random nature of DNA inheritance there will be
no hard and fast rule for pre-determining how useful the Ancestry by
DNA test will be, but my point was that if one has NO suspicion of
mixed ancestry (using the ancestral categories defined by DNAprint)
then the test is unlikely to uncover it, because any mixed ancestry
that did exist would quite likely be at the less than 5% level.

If one does have a suspicion of mixed ancestry the test could be
useful, but because of the relatively low number of markers tested
(69) the test can easily miss or miss-calculate percentages that are
not >>5%.


Peter Reed





This thread: