GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-05 > 1211397294


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] S21/S28 Split+m223 stuff
Date: Wed, 21 May 2008 13:14:54 -0600
References: <200805211855.m4LItKUq013470@mail.rootsweb.com>


----- Original Message -----
From: "Tim Janzen" <>
I summed the [latest]
> mutation rates for each version and found that the sum of the mutation
> rates
> for markers 38-67 from John's earlier set of rates was 0.04469 and that
> the
> sum of the mutation rates for markers 38-67 from John's most recent set of
> rates was 0.04222. Thus, there is about a 5.5% difference between the two
> sets of rates with John's earlier set of rates being slightly higher than
> John's most recent set of rates. This affects the age estimates generated
> by Ken's current program and causes them to be younger than they would be
> if
> Ken were using John's most recent set of rates.

[[[[ All that enters the unweighted marker case is sum of ALL mutation rates
being used. I mentioned before that sum is about .12 The 1 - 37 markers
are substantially faster on average than the 38-67 markers. So a change of
.00247 in the 38-67 panel amounts to a 2.1 percent change in the total
mutation rate being used, and a corresponding 2.1 percent change in
estimated generational times. So while welcoming getting the latest rates
inserted into the program, I could only wish to be worrying about 2 percent
issues in age estimates.

The most accurate rates possible will be important if the weighting issue I
recently discussed turns out to be significant in the overall product, as it
seems it will for deep time, especially for the few fast markers in the
38-67 panel.

Ken ]]]]]

Ken


This thread: