GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2007-02 > 1170487073
From: "John McEwan" <>
Subject: Re: [DNA] Article: Efficient inferenceonknownphylogenetictreesusingPoisson regression.[Human mtDNA]
Date: Sat, 3 Feb 2007 20:17:53 +1300
Perhaps I should not quote without the context. He was saying that the
demographic history results derived from the autosomes do not match
exactly with results from mitochondria where they can be compared (the
differences in mutation rates cause problems at some time depths).
If there are differences between the two, this can be due to various
causes. One such cause he proposed is positive selection sweeps in mtDNA
(basically a favourable change being selected for). If this did occur
then using mtDNA to age events (as you mention) and inferring human
migrations followed the same mtDNA pattern and timeframe is fraught.
Autosomes cross over and are not like mtDNA or the Y chromosome in that
regard. However, at very close distances this happens very rarely if at
all and such events can be detected, so these small regions sort of act
like mini versions of mtDNA (and these regions are often about as big as
mtDNA). There are a lot of them. You can date the emergence of the
haplotypes for each and build trees from them. It is these trees and
dates that generally do not match with mtDNA.
If you look at your reflection and the mirror is distorted, you will see
a distorted view. Statistics may refine the accuracy of what is observed
but it is still distorted unless you address the problem with the mirror
itself (in this case the violation of the assumptions used in the
He was not denying what has been observed, he was just saying you have
to be very careful not to over-extrapolate from the results obtained.