Archiver > GENEALOGY-DNA > 2003-11 > 1069787014

From: Charles <>
Subject: Re: [DNA] Measuring Genetic Distances
Date: Tue, 25 Nov 2003 14:03:34 -0500
References: <>
In-Reply-To: <>


Welcome to the List.

I for one will really enjoy reading the discussions and debate between
you and the mathematicians and statisticians on this List as to the
correct or most correct way to compute the concept of "Genetic
Distance", for use by genealogists in comparing Y-DNA test result
differences of two or more tested people in surname projects. There has
been much confusion in this area, in my opinion, over which way is best,
i.e., the "sum of the differences" method or the "sum of the squares of
the differences" method, possibly confusing newbies, since some in this
list challenge the "sum of the differences" way it is presented by
FamilyTreeDNA in their surname project group administrators "Genetic
Distance" report on their website. It is my understanding that
FamilyTreeDNA basically uses your methodology in their algorithm to
calculate Genetic Distance. That is why I asked for your assistance in
discussing this with Mr. John Chandler in this list forum regarding his
advocating the "sum of the squares of the differences" method in order
to clarify this issue, in the interest of establishing a consensus and
an "industry standard" definition of the correct way to calculate
"Genetic Distance" for use by genealogists.

I am glad you are taking to the time to create that website to help
clarify this issue. Thank you.



Bruce Walsh wrote:
> Greetings.
> Given all of the discussion about the appropriate method for genetic distances (squaring versus simply summing the total number of differences), I've posted some material on this at the following website:
> I have heard that this is an on-going topic (I've just signed up for a couple of weeks), with lots of excellent discussion, motivated by the correct idea that the actual number of changes underestimates the observed number of changes and hence some correction is appropriate.
> Bottom lines:
> • A genetic distance attempts to estimate the actual number of underlying mutations that have occurred from some observed difference, here (Y chromosome markers) in the allele length (or state)
> • The expressions for translating from an observed difference to an estimate of the actual number of mutations are complex and are also a function of the time to MRCA
> • The simple sums of difference method works for fairly recent sets of ancestors (up to 200 generations)
> • The sums of square differences method is inappropriate.
> I've yet to post the technical details (i.e., the actual equations), but will do so over the next few days. I have to testify at a Frye hearing within the next few hours to determine how the State of Arizona will use Y chromosome markers in Forensic tests. Mike Hammer, another ftDNA advisor, will also testify. Frye hearings are for the admissibility new scientific approaches. Mike will address the molecular genetic aspects, I'll be addressing the statistical and population genetic aspects. Interesting, Arizona has not yet used Y chromosome markers and Mike and I will serve as experts for the State in this case. (Note that ftDNA has no vested interest, one way or the other, in forensic applications, and is no way involved in any aspect of this trail other than having two of its scientific advisors testifying as members of the Faculty of the University of Arizona).
> Others hold differing views on genetic distance and I look forward to an interesting, engaging (and fun) discussion
> Cheers
> Bruce
> Bruce Walsh
> Associate Professor and Associate Department Head
> (Associate Editor, Genetics)
> Department of Ecology and Evolutionary Biology
> Biosciences West
> University of Arizona
> Tucson, AZ 85721 USA

This thread: