GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-02 > 1171522716


From: OrinWells <>
Subject: Re: [DNA] How do you do an estimated Common ancestor
Date: Wed, 14 Feb 2007 22:58:36 -0800
References: <c1e.100543a9.3305453f@aol.com>
In-Reply-To: <c1e.100543a9.3305453f@aol.com>


You are correct, I hit on the old calculator. But I remain skeptical
of the real-world value of this sort of predictor if people are
trying to use it as a predictor of when their common ancestor was
born. And the very name "Most Recent Common Ancestor" is probably
the reason it has been so misleading.. It just seems everything I
see does not behave in a nice predictable manner and I remain
convinced there are influences over the mutations that nobody
understands yet. Maybe we never will.

You are also correct that I do make an intuitive assessment of a
likely connection based on number of mutations. I have pretty much
come to the conclusion that even at 43 markers more than three
differences is probably enough to exclude the two from being closely
enough related that we are ever going to find the confirming
documentation. If they have a different surname I won't even
consider it at 3 differences unless there is solid evidence to
indicate an event that would have caused the surname change.

The DNA testing has great value for people with the same surname who
match each other with 0 to 3 mutations from what can be determined as
the likely modal value for the haplotype and who can find paper
trails that point to at least a common location where further
research can be launched. It takes off the table all the families
with that surname if they do not match. It can not exclude other
branches who appear on the scene independent of the ancestors that
are known. But genealogy is always a treasure hunt.

It can also be frustrating for those who don't match others either
because there is something we don't know about their family or where
some event has caused an unanticipated "alteration" of the DNA
somewhere up the tree. Life gives us hard lessons at times.


At 09:10 PM 2/14/2007, you wrote:
>In a message dated 2/14/2007 2:23:49 PM Pacific Standard Time,
> writes:
>
> > It CAN be expected that sons will differ from their fathers. It had
> > to happen somewhere for a mutation to be evident. What is VERY
> > unexpected is for two brothers to have haplotypes that not only
> > differ from each other but in each case differs from their father in
> > a different manner.
>
>I think you were using my MRCA calculator in your examples (not quoted here).
>That was written a long time ago, when 12 markers were the most you could
>order, and I only did calculations for a maximum of two mutations.
>The numbers
>are still correct, but there are other calculators out there now
>that will let
>you explore a wider range of possibilities. They would illustrate my "triage"
>statement better -- as the number of differences grows, there comes a point
>when you can pretty much rule out the two men being brothers. But
>when does that
>point come?
>
>I have another calculator,
>
>http://members.aol.com/dnafiler/MutationCalculator.exe
>
>which you can use to find the probability of observing two mutations in two
>brothers. I don't know what mutation rate RG is using for their set of 43
>markers, but let's say it is .003. Plugging in 43 for the number of
>markers, .003
>for the mutation rate, and 2 for the number of transmission events, the
>probabilities of observing a certain number of mutations are
>
>0 .772
>1 .199
>2 .026
>3 .002
>
>and so forth, with rapidly diminishing probabilities for 4, 5, 6+ mutations.
>
>So most brothers (77.2%) will match, but we would expect to see two mutations
>in brothers in about 2.6% of cases -- not "typical," but not what I'd call
>"VERY" unexpected either. In fact, as databases grow, we are bound
>to encounter
>such cases now and then.
>
>And if two mutations have indeed occurred in two brothers, it's far more
>likely that they would fall on different markers than on the same marker --
>there's only one chance in 43 that the second mutation will fall on
>the same marker,
>and 42/43 chance that it will fall on a different marker.
>
>No one is saying that TMRCA calculators help your target your research to a
>particular generation -- as you say, the 95% confidence intervals
>are very wide
>(and the last 4.9% is very very wide). But if you learn that there's only a
>.0001 chance of finding a common ancestor within 1000 years, you
>wouldn't want
>to invest much effort in that line of research (the triage concept). I'm sure
>you have a mental triage system in your project -- above a certain threshold
>of genetic distance for two people who claim to have a common
>ancestor, you're
>going to start suspecting that a non-paternity event or a laboratory error is
>coming into play. You're using the TMRCA concept qualitatively, even if you
>aren't using it quantitatively.
>
>Ann Turner
>
>
>
>
>
>-------------------------------
>To unsubscribe from the list, please send an email to
> with the word 'unsubscribe'
>without the quotes in the subject and the body of the message


Orin R. Wells
Wells Family Research Association
P. O. Box 5427
Kent, Washington 98064-5427
<>
http://www.wells.org
Subscribe to the "Wells-L" list on RootsWeb

This thread: