GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-02 > 1140995941


From: (John Chandler)
Subject: Re: [DNA] SNP Test Necessary: WEBB-BIRNBACH R1b SuperX HaplotypeStudy Report
Date: Sun, 26 Feb 2006 18:19:01 -0500 (EST)
References: <29.2cd36cd.312fb2d5@aol.com> <REME20060223225152@alum.mit.edu> <43FE94D7.4040207@sbcglobal.net> <REME20060224143303@alum.mit.edu> <4400B008.40404@sbcglobal.net>
In-Reply-To: <4400B008.40404@sbcglobal.net> (aburto@sbcglobal.net)


Al wrote:
> They appear to have a patent pending. Also they "published" their
> mutations rates in an Oct 2004 conference on genetic genealogy,
> somewhere. If there was a publication of those results (a conference
> proceedings), it sure would be nice to see it. FTDNA claims their
> mutation work is supported by the University of Arizona which is a
> respected organization. Even though they have not published in a peer
> reviewed journal, they appear to have a good foundation, good
> credentials. Also Dr. Walsh is on their staff. He is a a well respected
> scientist (I even ordered his book on genetics --- I just found he had
> written a textbook today).
>
> So, overall, summerizing things from my perspective, it is not clear to
> me that FTDNA is so far off base ...

I went over all this yesterday, but I guess you were hung up on the
spread of results obtained from direct mutation rate measurements
on individual markers. Now that you understand that point, you
need to read what I wrote then again, and it *will* be clear to you
that FTDNA is indeed very far off base. The direct measurements
give a result for the average rate of FTDNA's first panel of 0.002,
and the standard deviation in this measurement is +/- 10%. The
rate announced at the 2004 conference was 0.004, i.e., off by about
ten standard deviations from the direct value. There is a simple
word for a result like that -- "wrong". Even allowing for the
larger standard deviation in the FTDNA result, there is simply no
way to reconcile it with the published results.

Now, to address the specific points you mentioned above:

patent...
-- please be serious. That has nothing to do with the mutation rate study.

"published"...
-- please be serious. Announcing scientific results without any
scientific details is not publishing. Even the "cold fusion"
guys put out a preprint.

University of Arizona...
-- Yes, the mutation rate study was carried out at the U of A. FTDNA
supplied only the publicity in the form of a mass mailing to surname
project administrators asking them to volunteer, and announcements
of the pending study on their web page and newsletter.

Bruce Walsh...
-- is only peripherally connected to the study. As of last December,
he wasn't even aware of the recent father-son studies and thus had no
idea that they had blown the Arizona study out of the water. His
response to this news was to hope that something could be salvaged by
including both the published father-son data and the FTDNA
pedigree-based data in a combined analysis, but he didn't then have
(and, last I heard from him, still does not have) access to the
pedigree-based data. I leave you to draw your own conclusions
about how much involvement he's had in the study.

In any case, the problem with the FTDNA/Arizona study is that it
was based on a biased sample, and the only way to determine how
bad the bias was is to compare the final result with a known
calibration. In other words, the study yields information about
human psychology instead of information about mutation rates.
Since the calibration shows that the bias was indeed very bad,
we can't even be confident that the relative results between
panels are as reliable as the size of the study would predict.

John Chandler

This thread: