GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2002-04 > 1019669460


From: "John F. Chandler" <>
Subject: Re: [DNA] Relative Genetics/Ancestry.com results
Date: Wed, 24 Apr 2002 13:31 EDT


Brett wrote:
> This is an in-house number based on research...

"In-house" research that has not been published is always suspect.
The system of publishing results is not for the purpose of revealing
interesting facts -- it also allows reviewers to shoot down obvious
mistakes before publication and competitors to offer contrary evidence
afterwards. Here is just one example of a possible mistake that could
be affecting the "in-house" results: last year, the BYU studies
included a marker they called "E5" and which has since been revealed
to be the same as DYS310, which no other lab has used for genealogical
tests. It was clear from the results (assuming that they were, in fact,
correct) that the mutation rate for that marker is MUCH higher than for
any of the others. In a study that included descendants of a single
common ancestor, with 98 total meioses, there was no clear majority of
any one allele among the descendants, and five different alleles turned
up. Since we were not told the actual repeat counts for the arbitrary
"scores" in the report, it is impossible for me to determine the actual
number of mutations, but it had to be at least 6, and at least 2 of
those had to be double, and 1 triple! Thus, the estimated mutation rate
had to be at least 10% (+/- 3%). This is close to two orders of
magnitude higher than the "standard" rate and is so high that it no
longer qualifies as a "rare" event, which means that we should expect
some of the supposedly unmutated alleles to be the result of multiple
mutations that happen to cancel, thus driving the estimated rate even
higher. If any of the DYS310 test results are included in the
"in-house" figure, it is irrevocably tainted. Published results would
be expected to provide details like which loci are included.

Note: Allan mentioned the BYU world-wide research project involving
4-generation pedigrees and a target of 100,000 blood samples. None
of that project is relevant to determining the mutation rate, since
they are NOT going after coordinated sampling from closely related
individuals and NOT preserving names anyhow.

John Chandler


This thread: