GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2002-04 > 1019497260


From: "John F. Chandler" <>
Subject: Re: [DNA] Relative Genetics/Ancestry.com results
Date: Mon, 22 Apr 2002 13:41 EDT
In-Reply-To: glenneil@bellatlantic.net message <3CC44D1B.BCAA77E0@bellatlantic.net> of Mon, 22 Apr 2002 11:32:13 -0600


Clarke wrote:
> While my Glennon family Y-DNA project is at FTDNA, I did get my personal
> sample tested by Rel.Gen./Ancestry just to make a comparison and to get
> some results on additional markers...
> All of the markers tested by both labs matched EXCEPT for DYS 388.
> FTDNA reported 12 on that marker and Rel.Gen./Ancestry reported a 13.

Any disagreement is disturbing, given that the RelGen numeric results
are now supposed to be calibrated properly. The only other post-cal
report of a comparison between FTDNA and RelGen showed complete
agreement. Hence, we know that somebody goofed somewhere. It might
be a good idea to recall the error rate published by Wilson et al. --
they tested many samples twice independently and compared the results
to obtain an estimated error rate of 0.7%.

> one thing that is worth pointing out under the heading of "Most
> Recent Common Ancestor" that raised some questions in my mind based on
> statistics that I have seen earlier. It is:
>
> # of Mismatches - 0
> Median Number of Generations - 3.6
> Possible Range in 95% Interval - 0.1 to 19.2

If this is based on 24 markers and a mutation rate of 0.2%, then the
median should be twice as much: 7.2 generations, with a 95% confidence
interval of 0.5 to 31.2 (I calculated that with a binomial distribution
by doing a continuous integral instead of a discrete sum, but the error
should be small). Perhaps they are assuming a mutation rate of 0.4%?

> # of Mismatches - 1
> Median Number of Generations - 8.9
> Possible Range in 95% Interval - 1.3 to 29.7

Again, too small. I get 17.5 median or (3.7 to 49.4). Again, this might
be the result of assuming a higher mutation rate. Then again, maybe
they took a table of results that was based on MRCA and thought it was
instead based on mutations between one ancestor and one descendant (and
so divided by two an extra time).

John Chandler

This thread: