GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-09 > 1220985767


From: (John Chandler)
Subject: Re: [DNA] 9RA autosomal Native American marker
Date: Tue, 9 Sep 2008 14:42:47 -0400 (EDT)
References: <BLU126-DAV8BE82C64F3E07C5644BC792540@phx.gbl><ea3bd9560809090857y6dbab028u10debbf7a09819e1@mail.gmail.com>
In-Reply-To: <ea3bd9560809090857y6dbab028u10debbf7a09819e1@mail.gmail.com>(fauxdk@gmail.com)


David wrote:
> Not exactly Ray. It all hinges on how far back the Native American ancestor
> was in generations.

Everything hinges on that, including the DNAprint test results. What
Ray is proposing is an independent check on DNAprint via one particular
autosomal STR marker which apparently has a high frequency contrast for
one allele between populations. The problem is that it's only one marker
and so doesn't have very good sensitivity to the admixture and so doesn't
make a very sensitive independent check.

To be specific: if you take a collection of 100 individuals with an
average of 10% Amerindian autosomal DNA, they have a total of about 20
copies of Amerindian-origin D9S919 among them. Thus, they have an
expectation of about 20x0.37 = 7.4 instances of D9S919=9, and the
standard deviation is sqrt(7.4), or almost 3. Thus, the 95%
confidence interval is not restricted closely near 7.4, but instead
ranges from less than 2 to almost 13.

In other words, a count of 7 instances would not be a stunning
confirmation of DNAprint, and a count of 2 instances would not be
a stunning condemnation.

Indeed, the supposed frequency of 37% is perhaps not yet firmly
established over the whole range of Amerindians, and so the
95% confidence interval should perhaps be stretched even wider.

John Chandler

This thread: