GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-12 > 1260736355


From: "Tim Janzen" <>
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent
Date: Sun, 13 Dec 2009 12:32:35 -0800
In-Reply-To: <DF996FBD-C13B-4DE1-9ECC-376E929AE952@vizachero.com>


Dear Vince,
You make excellent points. One of my main points in recent messages
(and similar messages this past summer) is that one can't include a lot of
fast mutating markers in TMRCA for old haplogroups and subclades
(particularly if older than 20,000 years or so) before saturation of
variance for the fast mutating markers starts to skew the ages to be younger
than they should be. Ken is indeed correct that no one has measured
(quantified) the degree of saturation of the markers and that I don't know
how much saturation there is. However, you and I have enough insight to
realize that calculations that include large number of fast mutating markers
consistently produce TMRCA estimates that are inconsistent with well
established dates from archeology (the settling of Australia, etc). I
realize that the confidence intervals are relatively high for TMRCA
estimates that use only the 24 slowest markers, but at least you and I are
not using any arbitrary fudge factors like Zhivotovsky does. I can live
with higher confidence intervals if the results suggest that my TMRCA
estimates are likely to close to the true TMRCA.
I see no reason to "throw the baby out with the bath water". TMRCA
estimates from STR haplotypes are one of the main tools we have to available
to estimate the ages of haplogroups and subclades. I think it is better to
acknowledge the problems with associated with this technique than to simply
abandon use of the technique altogether.
Sincerely,
Tim

-----Original Message-----
From:
[mailto:] On Behalf Of Vincent Vizachero
Sent: Sunday, December 13, 2009 12:15 PM
To:
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent

I think it bears pointing out the advantage of adjusting the marker
set (the approach Tim and I use) as opposed to adopting an "effective"
rate approach (a la Zhiv et al.).

The problem Tim and I are solving (or trying to) is mutational
saturation: the fact that over longer time frames some markers
accumulate variance in a non-linear manner due to range constraints.
The Zhiv approach is to apply a "factor" to the real mutation rate,
but that still leaves you in a position of trying to estimate a non-
linear curve with a linear equation.

Ken is right to observe that right now we are engaging in a bit of
circular reasoning, because right now the only way to know which
markers to leave out is to "know" what kind of TMRCA estimate you
should end up with. I think we have bootstrapped it as well as can be
done using independent estimates for the TMRCA of Y-Adam and other
major nodes (e.g. from Karafet et al.), but I do agree with Ken that
we'd much rather have a less circular approach. And a whole lot more
markers with the right parameters so our confidence intervals weren't
so huge.

But since Zhiv et al. were ignoring some pretty glaring assumptions,
the approach that Tim has described is certainly far more nuanced than
the one advocated by Zhiv et al.

VV



This thread: