Archiver > GENEALOGY-DNA > 2003-11 > 1069273306

Subject: Re: [DNA] Marker DYS390 in World database
Date: Wed, 19 Nov 2003 15:21:52 -0500 (EST)
References: <004a01c3ae58$5775f850$8a0aa8c0@William>
In-Reply-To: <004a01c3ae58$5775f850$8a0aa8c0@William>(

Bill wrote:
> Have you determined if geographic location, or any other natural phenomenon,
> is causing the variation in the mutation rates on the same NRY DYS390 loci
> in the different branches of the Wells family.....??? For example, if I am
> a Wells and move from an area where the mutation rate is lower to an area
> when the Well branches have higher mutation rates.....will my offspring have
> great mutation rates of DYS390....???

Unfortunately, we are in the zone of small-number statistics. It
might help to look at the probabilities for a "typical" surname study.
Let's suppose the study has covered 20 different families, with
results for 25 markers and a total of 30 mutation opportunities in
each family. (That's enough for four centuries of history if there
are only two testees in the subgroup.) At the nominal mutation rate,
the chances of seeing NO mutations in a particular subgoup are about
1/4. I.e., we would expext to find roughly five lines in the study
that have been "stable" over four centuries. The chances of NOT
finding any such stable lines are less than 1%. At the same time, the
chances of finding four or more mutations in any particular subgroup
are about 1/16. I.e., we would similarly expect to find roughtly one
line with at least four mutations in evidence in that same time
period. The chances of NOT finding any such high-mutation groups are
only 1/4.

Note that the average mutation rate for one of these high-mutation
groups would be at least 0.005, "substantially" more than the accepted
nominal average rate, but this apparent "violation" of the average is
nothing but a statistical fluctuation. We EXPECT to see statistical
fluctuations whenever there are statistics. It's just that the
fluctuations tend to get smaller as we accumulate more data --
provided that we average all the data together. If, instead, we focus
on the differences from group to group, then the fluctuations actually
get LARGER as we include more groups.

John Chandler

This thread: