Archiver > GENEALOGY-DNA > 2006-01 > 1138418113

From: (John Chandler)
Subject: Re: [DNA] autosomal testing
Date: Fri, 27 Jan 2006 22:15:13 -0500 (EST)
References: <068801c62379$3c17de50$0101a8c0@HighReaches.local>
In-Reply-To: <068801c62379$3c17de50$0101a8c0@HighReaches.local>(

Glen wrote:
> Resolution is low, but I think that that has more to do with the
> current limited size of the database than anything else.

DNAprint is not like the FTDNA haplogroup estimator. It does *not*
work with a database of individual test results. Instead, it uses a
set of allele frequencies averaged over a reference sample from each
target population. In that sense, it is somewhat like Whit Athey's
haplogroup estimator. The issue is not the *size* of the samples,
but rather their *representativeness*. Making the samples larger
would not help unless the added test results were designed to make
each sample more representative of its target population.

> I'll leave it to one of the geneticists to address this in detail, but while
> small percentages are problematic due to standard gene mixing, there would
> at least be a chance of significant indicators showing up.

The hallmark of these markers is that only a very few are significant
indicators by themselves, and each one is only good for its share
of the whole set of markers (i.e., a swing of about 1% on the 2.0
version of the test of about 1/2 of 1% on the 2.5 version). All
the rest of the markers are just moderately more frequent in one
population than another.

John Chandler

This thread: