GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-07 > 1183553958


From: "Joe Knapp" <>
Subject: Re: [DNA] Help please with marker 458
Date: Wed, 4 Jul 2007 08:59:18 -0400
References: <008701c7b415$794b0120$2401a8c0@your447023ae6b><REME20070625220955@alum.mit.edu><a81622ac0706260342k14ceb7e4gfe92491fbc966d2b@mail.gmail.com><REME20070626174750@alum.mit.edu><a81622ac0706261718v542d02e8nf80d7ed4c9ccf2b9@mail.gmail.com><REME20070627010107@alum.mit.edu><a81622ac0706270842g1475445fs92318768d6443227@mail.gmail.com><REME20070629172639@alum.mit.edu><a81622ac0706291814v778c80ccl9882806072865762@mail.gmail.com><REME20070703220030@alum.mit.edu>
In-Reply-To: <REME20070703220030@alum.mit.edu>


On 7/3/07, John Chandler <> wrote:
> > http://coolohio.com/dna/hapdist.jpg
>
> There is a problem with that histogram. The 37/37 bin appears to
> have about 50,000 pairs, but the count should be only about 20,000.
> Also, the shape is very different from what I see for the two
> 29-marker subsets, which closely resemble each other.

I'll check that... I just hacked together that histogram software the
other day--there may be a bug.

> Speaking of the histogram, I note that it is completely different
> in character from a real-life sample of FTDNA customers. The
> latter are the product of a 10,000-year population explosion,
> while your simulation experienced rigid ZPG. Judging by the
> very erratic mutation model curves, I would need a lot more
> haplotypes to smooth out the results. Nonetheless, I might as
> well give you the results for this batch. The most obvious
> thing about this set of results is that the range of calculated
> rates is much wider than the input range you mentioned above.

I'll look at this more closely later, after July 4th activities &
such. At first glance, it looks like your results generally track the
actual mutation rates, although there are a few glaring exceptions as
you note. I think quite a few of them are outside three-sigma limits.
I could do another run later with more haplotypes.

Joe

This thread: