GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2011-11 > 1321176823


From: "Lancaster-Boon" <>
Subject: [DNA] Correct TMRCA analysis (Europe)
Date: Sun, 13 Nov 2011 10:33:43 +0100


Because there are lots of mails about the same thread, I'll write a combined
comment. By the way I second the recent call for better formatting of posts!


Gary Felix: "How does an individual come into a population in Europe and
proceed to replace it 6K years ago? Possible with some sort of genetic
advantage if that is what you are writing about."

I do not believe it requires a genetic advantage, even though the effect is
similar. Y DNA does not mix, so the different types of Y DNA in Europe are
like oil and water. Every generation in every family one type pushes out
another type. So by necessity they come and go relatively quickly. (And are
not especially useful for understanding full ethnic ancestry.)

Gary Felix: "How are people coming up with these old ages for the V13 V12
node with your [Ken's] utility?"

Actually for the V13-V12 node, everyone comes up with those sort of ages, or
older? Anatole complains in fact that it is almost useless, but I think his
remarks are not correct either...

Anatole: "The interclade between V12 and V13 shows a time span to A COMMON
ANCESTOR of V12 and V13, that is E-M78. [...] Why to stretch the TMRCA for
V13 down to M78?? [...] Absurd. This is the way to kill any work. Make it a
mess and claim that nothing works."

Actually this remark has some aburdity to it. :) E-V13 is also E-M78, and so
tracing back to a common ancestor who was E-M78 is exactly what we want to
do if we wish to discuss the age of E-V13 as such (the first mutation) as
opposed to the most recent common ancestor of all modern E-V13. Given his
strong remark though, Anatole's remark seems indicate that Anatole thinks
the common ancestor of E-V13 and E-V12 will be way back near the origins of
E-M78 itself, which is absolutely not correct. E-M78 itself once underwent
expansion, which started in Africa, and E-V12 and E-V13 are extremely likely
to have a much more recent common ancestor than that. (Before the discovery
of V12 and V13, STR clustering showed that some men look at bit like both -
in particular a small set of Druze in the Middle East.) Anatole seems to
have misread the context of the remark he answered and written too quickly.

I would like to thank Lawrence for his quick calculation. I would be
interested to see other interclade checks such as V22 and V13.

Ken: "I, unlike some of the stick-in-the-muds, actually think the SNPs with
higher mutation rates, which some call "unstable", are more informative and
powerful for our hobby right now than the "slow" SNPs."

I agree and as a genealogist I would say it is even more important. FT DNA's
last two new panels were disappointing for anyone interested in genealogy,
but seem to have been selected partly on the basis of an idea that this
would suit the assumptions of paying customers who are interested in trying
to be population geneticists. Problem is that even for that purpose, slow
markers essentially only help us pick the low hanging fruit which has
already been discovered anyway.

FT DNA has more fast markers, ever since they took over Thomas Krahn's
company. But these markers are even more difficult to use on their website
than the 111 marker new panel. This is an issue which really should be
addressed. FT DNA should be out there explaining how great those advanced
markers are for genealogy.

David Faux: "not only are Y-STR MRCA analyses for "deep ancestry" on their
"way out" due to "poor and erratic performance" (as per Dienekes
assessment), but the same can be said for their use for genealogical
timeframes"

A good and challening point. But, first I would say STR analysis is still
absolutely fine for cluster analysis, both for long term family trees and
recent ones. The problem described by Dienekes is using STRs for time
estimations, like a "clock". Second, the reason we are getting such tiny
returns on the new panels from FT DNA is clear to me: the markers selected
are all ones which mutate very infruently.

The best way to use STR haplotypes is clustering. That is where Ken has also
put a lot of effort, and Anatole's repeated warnings about finding the right
"bases" also seems connected to this. We keep seeing that STR clustering
does work very well to predict SNPs whenever we find new SNPs to cross
check.

But here is a problem for genealogy: in genealogy, clustering is the
"triangulation" we can do within surname projects, to try to see who shares
more than one mutation. It is, or would be, the best way to use DNA in MOST
genealogical questions. But time after time I see people with 67 markers,
and common ancestors centuries ago, whose haplotypes are SO CLOSE that they
do not even share two mutations. The STR markers in the 67 and 111 marker
upgrades are two "stable".

In contrast, when the 37 marker upgrade came out, it led to many
genealogical breakthroughs. (But many people complained about the fact that
it had too many fast moving markers, apparently basing their negative
feelings on the wrong-headed idea that the best hope for genetic genealogy
was to find a highly accurate molecular clock. This project has indeed
failed.)

The advanced markers are, or would be, much more useful for clustering or
triangulation within a genealogical timeframe, but hardly anyone has tested
them, and we can not compare to others without enormous difficulty.

Didier: "I still have doubt on the V13 result (the STR discussion below
doesn't prove the V13 case).

I think it must be correct. The STRs agree with the SNP result, for example
if you look at the relatively close matches on the modern E-M35 project data
set. What are the chances of such a coincidence?

Didier: "If both groups were V13+ with a bottleneck, one would expect many
more SNPs separating the "new" V13 group from other E-M78 subclades."

They only tested for critical known SNPs. They did not do sequencing. As it
happens we do know a few SNPs which should be equivalent to V13, and so the
authors could check for those.

Anatole: "The 7,000 ybp excavated fellow was NOT a common ancestor of our
contemporary V13 folks. It has a 4-mutation-off haplotype, it is distant by
3750 years, and our contemporary V13 are NOT his descendants."

I think it is impossible to be so confident of this. Whatever a statistical
study indicates, we have quite close modern matches in the E-M35 project.

Best Regards
Andrew

This thread: