Archiver > GENEALOGY-DNA > 2007-03 > 1174690021

From: (John Chandler)
Subject: Re: [DNA] Cruciani and 2007 TMRCA estimates
Date: Fri, 23 Mar 2007 18:47:01 -0400 (EDT)
References: <>
In-Reply-To: <>

Elizabeth wrote:
> While #3 seems to be based on 'real'
> statistics, I wonder how many father-son comparisons were made to arrive at
> those rates.

You seem to be suggesting that the many published father-son studies
are based on fake data. What basis do you have for that suggestion?
Putting the published studies together, the database consists of 6,000
to 10,000 father-son pairs for each of the "core" loci (DYS385,
DYS389i, DYS389ii, DYS390, and so on). The collections range on down
to 50 pairs for DYS413. In all, I've seen reports on 28 different
Y-STR loci, with a total of 98,421 meioses.

> I wonder how Forster and Zhivostovsky arrived at their rates.

You may find it helpful to read their papers.

> I wonder what years/generations was used and what SD applies to any of these
> rates.

Zhivotovsky et al gave a standard deviation for their estimated rate,
but it wasn't really applicable, except as an uncertainty in the
"effective" rate for the specific populations they studied. The
systematic uncertainty would be much higher in trying to transfer that
rate to some *other* population with a *different* population history.

> Partly because my DNA project is with Family Tree, and partly because I
> suspect a good number of other group administrators who follow this list
> also are based there, I must continue to wonder why so many consider FT's
> mutation rates as too short, since the database they have available to them
> in determining the rates they use is probably much larger than other
> studies - maybe I'm wrong.

Yes, you're wrong about that. The rates used in FTDNATiP are based on
a study carried out several years ago, but never published. It was
based on pedigrees submitted to the study by project administrators.
The response rate was not as high as one would hope. The only details
released are the average rates for the three panels available from
FTDNA at that time. The 12-marker panel, in particular, is well
covered by direct father-son studies. The FTDNA rate is too high by
more than a factor of two. There is no question about that.

> I don't know why they don't publish their
> results - I'm asked about it a number of times.

Wouldn't you say that knowing the answers are wrong is a deterrant
to publishing them?

> It appears most Group Admin's - at FT anyway - use the FT TiP calculator
> when compiling results for their projects. What credence can we provide in
> reporting to our participants the results we are producing, when the larger
> community suggests their calculations are inadequate?

The method of calculation used by FTDNATiP is ok. The only problem is
the assumed rates. You can scale the times to compensate and have the
best of both worlds.

> I have reviewed (mainly Irish/Scottish, I admit) are seeking answers to the
> origin myths of their ancestors - while they are usually still A.D., we may
> be going back 2000+ years.

If you are talking about the coalescence time of a whole population,
rather than the TMRCA of two individuals, then the whole range of
issues about population dynamics, sample biases, and sample sizes come
into play. You surely don't have to worry about natural selection on
this time scale for Y-STRs, but I think that's the only worry you can

John Chandler

This thread: