GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-02 > 1171605677


From: Vincent Vizachero <>
Subject: Re: [DNA] TMRCA
Date: Fri, 16 Feb 2007 00:01:17 -0600
References: <KHEKIJEABJGJEKDPFEDMIEOEDDAA.elizabethod@eircom.net><1132CF4A-B306-404E-9C42-F55BBB0F76FF@vizachero.com><001001c75177$0aa1c340$6400a8c0@Ken1><D850E775-C365-47FA-8E2F-800637BAF8ED@vizachero.com><002a01c7517d$d108b880$6400a8c0@Ken1>
In-Reply-To: <002a01c7517d$d108b880$6400a8c0@Ken1>


On Feb 15, 2007, at 9:52 PM, Ken Nordtvedt wrote:

> But that's just a claim; how about an explanation with a bit of
> science as
> to why?

Ken,

Lots of folks a lot smarter than I am have tackled this one (e.g
http://abc.zoo.ox.ac.uk/Papers/tig06_clocks.pdf).

Zhivotovsky make a big deal of bottlenecks and population size
concerns, and I think there is some validity to that argument even if
it is often overplayed.

One important factor less frequently discussed is mutational
saturation. STRs typically do not have an infinite allele range, and
thus over some period of time they lose their ability to convey
phylogenetic information. The signal effectively dissolves into
noise, and it does so at an increasing rate as the timescale
increases. In many cases, this happens in hundreds of generations
(less than 10,000 years).

Another factor is the risk inherent in extrapolation. This quote is
from H.-J. Bandelt et al. They are speaking specifically about mtDNA,
but the principle is transferable, I think.

> Nucleic Acids and Molecular Biology, Vol. 18
> Hans-Jürgen Bandelt, Vincent Macaulay, Martin Richards (Eds.)
> Human Mitochondrial DNA and the Evolution of Homo sapiens
> © Springer-Verlag Berlin Heidelberg 2006
>
>
> As for point 2, one has to bear in mind that the rate is being
> measured
> by single-generation events but the typical application involves
> time frames
> of 1000 generations or more—in other words, extreme extrapolation
> is employed.
> Ho et al. (2005) are right—although for another reason—when they
> contend that “it is invalid to extrapolate molecular rates of
> change across different
> evolutionary timescales”. Among the reasons that we would envision
> in the first place are sequencing artefacts, which are deemed to be
> unavoidable
> (according to Herrnstadt et al. 2003; Chap. 6). Suppose we had a new
> sample of 560 pairs of control-region sequences from mother–
> offspring pairs.
> In a worst-case scenario, we could have, say, 1.6 phantom
> mutational events
> per pair on average (extrapolating the conservative error estimate
> from the
> data of Nasidze and Stoneking 2001; Chap. 6). If we diluted and
> dispersed
> such error-loaded sequences among ideal sequences, so that only
> approximately
> 1% of the sequences were not ideal and carried that specific error
> load,
> then we would infer a ‘pedigree rate’ that is more than sixfold
> higher than
> the conventional rate.

Vince

This thread: