GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-08 > 1154546303


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] research strategy for genealogists
Date: Wed, 2 Aug 2006 13:18:23 -0600
References: <545.2630d300.32022165@aol.com> <6.0.1.1.2.20060802130532.03fbf880@mail.utexas.edu>


What are the two haplotypes converging from? If they are of the same
haplogroup or clade or variety, they both descend from a common founder who
existed at some time in the past. They don't need to converge; they may
have wondered around by mutations since the founding along similar paths.

If the two haplotypes are from different haplogroups, and we are talking
haplotypes with a decent number of markers, the probability of converging to
very similar structures is incredibly small.

Would someone who champions the notion of "haplotype convergence" please
explain what they are talking about?

Ken
----- Original Message -----
From: "O. W. Odom" <>
To: <>
Sent: Wednesday, August 02, 2006 12:48 PM
Subject: Re: [DNA] research strategy for genealogists


> Ann,
>
> Your message of 8/2/2006, copied below, seems to be addressing the
> likelihood of random convergence, showing that random convergence to a
> mismatch of 4
> (out of 37) is much more likely than random convergence to a mismatch of
> 1, which in turn is 74 times as likely as random convergence to a perfect
> match. Are you saying that a mismatch of 4 with a different surname is
> most likely explained by random convergence? In your earlier message, also
> copied below, you say that random convergence to a perfect match with
> another surname is much less likely than a MRCA before surnames were
> adopted. So somewhere between a mismatch of zero with another surname and
> a mismatch of 4 with another surname the most probable explanation has
> swung in favor of random convergence rather than pre-surname MRCA? Is this
> an accurate inference and, if so, where between zero and 4 mismatches with
> another surname do you think the pendulum would swing in favor of random
> convergence?
>
> Also, thank you so much for your many valuable contributions to the list!
>
> Obed
>
>
>
>
> At 10:40 AM 8/2/2006, Ann Turner wrote:
>
>>2) Is a 33/37 match with people of different surnames, but whose ancestry
>>goes back to the same vicinity & timeframe, worth pursuing?
>>
>>Not in my view. The reason is that allowing so many mismatches widens the
>>pool of candidates enormously. There are 74 different ways to differ by
>>one step
>>(one more or one less repeat at each marker). There are 2,738 ways to
>>differ
>>by two steps: each of those 74 haplotypes could differ in 74 different
>>ways,
>>but half of those would revert back to the original
>>
>>74*74/2 = 2,738
>>
>>Doug McDonald posted the general formula for J mismatches in N markers in
>>his
>>response to my post (scroll down in my post to get his URL):
>>
>>http://archiver.rootsweb.com/th/read/GENEALOGY-DNA/2004-09/1095511528
>>
>>It's (2*N)^J/J! (2 times N raised to the Jth power divided by J
>>factorial)
>>
>>For a 33/37 match, that's
>>
>>(2*37)^4/4! = 74^4/(4*3*2*1) = 1,249,441 distinct ways of differing by 4
>>markers on a 37 marker test.
>
>
> On 8/1/2006 Ann wrote:
>
> It is surprisingly difficult to achieve this state of
> coincidence/convergence. I know you meant it to dramatize your point, but
> your specific example would
> be well nigh impossible to achieve. I ran some simulations one time with
> easier starting points -- just one or two differences out of 37 markers.
> The
> results were summarized in this old message (long URL, ends in +F)
>
>
> http://listsearches.rootsweb.com/cgi-bin/ifetch2?/u1/textindices/G/GENEALOGY-DNA+2005+17283828476+F
>
> Convergence occurred in a very small percentage of cases, so it's not a
> good
> explanation for why different surnames have the same haplotype. Relatively
> speaking (pun!), it's far more likely that they have the same haplotype
> because
> they descend from a common ancestor who lived before the time surnames
> were
> adopted. It's harder to estimate the relative probability of another
> explanation,
> that different surnames have the same haplotype because of name changes
> and
> misattributed paternity in more recent times, where there's some chance of
> finding a paper trail.
>
>
> ==============================
> Search the US Census Collection. Over 140 million records added in the
> last 12 months. Largest online collection in the world. Learn more:
> http://www.ancestry.com/s13965/rd.ashx
>
>



This thread: