GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-12 > 1260305958


From: "Alister John Marsh" <>
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent
Date: Wed, 9 Dec 2009 09:59:18 +1300
References: <mailman.279.1260259242.26095.genealogy-dna@rootsweb.com><A93225B377724B83BF5250F368878955@anatoldesktop>
In-Reply-To: <A93225B377724B83BF5250F368878955@anatoldesktop>


Anatole,

You said...

>>>>>>>
Folks who work with mutations in haplotypes keep working and analyzing new
and new data.
And they come again and again to the conclusion that STR mutations have the
same frequency
now and thousands and millions years ago. There is not a single proven fact
or an observation
which would show otherwise. Recently a paper was published (Sun, J.X.,
Millikin, J.C.,
Patterson, N., Reich, D.E. 2009. Microsatellites are molecular clocks that
supports accurate
inferences about history. Mol. Biol. Evol. 26, 1017-1027) which showed that
mutations are
ticking with a constant rate for the last two million years at least.
<<<<<<<

I suspect that STR mutations have relatively similar mutation behaviour now
to millions of years ago.

But that does not mean that STRs mutate randomly, with each mutation event
having 50% chance of being a mutation up or down. It could be that for some
STRs, there is a favoured "sweet spot" which mutations tend to hover around.
DYS439 is moderately fast mutating, but seems to predominantly hover around
a narrow range of values.

Uniform mutation behaviour does not mean that some STR values are not
favoured for long term survival over others. It does not mean that an
allele value which has appeared benign for hundreds, thousands, or millions
of years may not be fatal or highly favoured in an extraordinary population
stress event which causes a "bottle neck".

You have confidence in statistical analysis of mutations. However, when the
haplogroup aging by variance of variance method was discussed some time ago,
it emerged that some doing the analysis selectively culled the data,
removing closely related haplotypes, and removing "outliers" which seemed
exceptions to the preconceived rules of mutation. Further, some selectively
culled the markers to use in calculations based on various factors including
mutation rates. Using either slow or fast mutating markers in calculations
enabled differing results to be obtained which varied by a factor of 50
fold.

I am not a mathematical expert. But it just seems to me that some
assumptions are underpinning results from statistical analysis which have
not been fully tested, because they are hard to test with available data.

As we keep saying, if we can recover good Y-DNA from well dated
archaeological sites, it would give some good data to verify the reliability
of statistical analysis of mutations.

To a significant degree, the statistical analysis lobby group have changed
my views on ages of haplogroups, particularly younger subclades, but I would
feel more comfortable with further corroborating evidence for estimates
involving longer time spans.

Some genealogists are happy to accept a birth certificate as
incontrovertible evidence of parenthood. Other genealogists like to
corroborate evidence by evidence from more than one source. I make no
apology for leanings towards the conservative group who seek more than one
piece of evidence.

When I became involved in genetic genealogy, I proved that two well
documented lines, regarded for 400 years as descending from two brothers, in
fact have male lines from different haplogroups.

Anatole, your confidence in the statistical analysis may be well placed. You
may be right. I am just one of those painful people who is sitting on the
barbed wire fence until more evidence is available. With advances in DNA
science, it is likely we will have good archaeological DNA evidence before
long. I hope!

John.






This thread: