GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-05 > 1211495068


From: "Tim Janzen" <>
Subject: Re: [DNA] S21/S28 Split+m223 stuff
Date: Thu, 22 May 2008 15:24:28 -0700
In-Reply-To: <0A16794D-21A8-4A8B-9B35-4032F2C536BF@vizachero.com>


Dear Vince,
The calculations for the TMRCA for various subclades of R1b that I
posted last night were not done using Ken's program or using one from James
Heald. I used my own spreadsheet. See http://www.timjanzen.com/dna.html)
for the basic spreadsheet. I announced this on the list several weeks ago
(see
http://archiver.rootsweb.ancestry.com/th/read/GENEALOGY-DNA/2008-05/12098797
01). I used the variance method that Ken has suggested in the past in
developing the spreadsheet on my web site. See
http://archiver.rootsweb.ancestry.com/th/read/GENEALOGY-DNA/2008-02/12030075
81 for one of Ken's first postings on the list about this method. For a
description of James Heald's method see
http://archiver.rootsweb.ancestry.com/th/read/GENEALOGY-DNA/2008-05/12100813
70. The ages for the column "KN +M" in my message last night come from
lines 3281 to 3296 in that spreadsheet. The ages found in the other columns
I posted last night came from the output of additional options I have added
to the spreadsheet since I originally put it on my web site.
I will try to get an updated version of my spreadsheet on my web site in the
next day or so.
Several days ago I downloaded an earlier version of Ken's
spreadsheet from his web site and created a new section in my spreadsheet
that used his ASD formulas. See
http://knordtvedt.home.bresnan.net/GenBackup.xls for his current
spreadsheet. After adding back the markers that he has omitted from his
spreadsheet the output in generations and in years was identical between his
spreadsheet and mine. This isn't surprising since we were both calculating
the variance, but we were using different methods to get there.
As Ken mentioned in his last message, neither of us have weighted
the markers in our spreadsheets. I am currently working on helping Ken with
the issue of weighting the markers.
As I have mentioned previously we need to interpret these results
with caution. There are number of factors that influence these age
calculations including the mutation rates that are used and the diversity of
the haplotypes used in the calculations. As both you and I have mentioned
in previous posts, a large number of haplotypes from relatively closely
related people skews the ages so that the TMRCA looks younger than is
actually the case. Thus it is possible that a smaller group of selected
haplotypes that represent the various varieties within a subclade may in the
long run prove to be more accurate than simply packing the spreadsheet with
large numbers of haplotypes. We need to do more testing on this situation
in the future.

Sincerely,
Tim


-----Original Message-----
From:
[mailto:] On Behalf Of Vincent Vizachero
Sent: Thursday, May 22, 2008 12:22 PM
To:
Subject: Re: [DNA] S21/S28 Split+m223 stuff

The ratio of estimated TMRCA for S21 and S28 depends on the way in
which the markers are weighted and which markers are considered.

I've copied an excerpt from Tim Janzen's post yesterday in which he
compared the datasets using your tool and using James Heald's tool.

In some cases (the 37 marker cases, for example) S21 actually looks
slightly older than S28 using your tool but dramatically younger
using James'.



On May 21, 2008, at 8:11 PM, Tim Janzen wrote:

> Project JH -M KN -M JH +M KN +M # of haplotypes # of markers
>
> R1b1c9 S21+ 4110 3486 4050 3539 190 12
> R1b1c9 S21+ 4808 3465 4224 2727 180 37
> R1b1c9 S21+ 3349 3075 3462 2431 120 67
>
> R1b1c10 Kerchner 6655 4001 6937 4876 62 12
> R1b1c10 Kerchner 5928 3352 5401 2537 59 37
> R1b1c10 Kerchner 6224 3589 5924 2850 43 67



This thread: