GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-02 > 1139839072


From: Doug McDonald <>
Subject: Re: [DNA] Sorenson Marker Details
Date: Mon, 13 Feb 2006 07:57:52 -0600
References: <43EFF12F.1090807@comcast.net>
In-Reply-To: <43EFF12F.1090807@comcast.net>


Bonnie Schrack wrote:
> Hi Doug,
>
> I remember you are the one who gave me the formula last year. I was
> wondering if you were the only one who had ever come up with a formula
> for this.
>
>> You don't multiply, you divide.
>>
>> I suggest using 0.02/rate to get weights.
>>
>> I use these values, in FTDNA order, for the 37 FTDNA markers:
>>
>> 22
>> 6
>> 14
>> 7
>> ...<snipped>
>
> But Doug, if those are the weights you're using, you seem to be using a
> really different set of mutation rates from the ones presented in Dean's
> tables.
>
> Could you give an example of the mutation rates for those first few
> markers, that you're using to get those weights?
>


I have numerous "sets" of numbers. One is just the numbers from the
Sorenson paper, normalized to my own average of the YHRD numbers and
the Kerchner numbers, all properly weighted by number of instances,
for the first FTDNA 12. The second is my own numbers, identically
weighted, taken from data from Ysearch, using a properly weighted,
for all Europe, set of data from R1b, R1a, I, and E3b. The third
is just the average of those two sets, where they overlap.

For the 37 FTDNA numbers these values are (FTDNA order):

SorensonMine Average

0.0009513310.0008590770.000905204
0.0037373710.0027890680.003263219
0.0013165740.0015893130.001452943
0.0026841120.0032557420.002969927
0.0023783270.0012155560.001796941
0.0023783270.0032427450.002810536
0.0002293390.0002818800.00025561
0.0003227730.0004284460.00037561
0.0035505020.0049408020.004245652
0.0018516970.0022469610.002049329
0.0012656100.0009810380.001123324
0.0021914580.0026317450.002411602
0.0049299320.0056478660.005288899
0.0010022950.0006891370.000845716
0.0010022950.0017997930.001401044
0.0002650140.0002873630.000276189
0.0001996108.66819E-050.000143146
0.0033050250.0032265960.003265811
0.0014907010.0014042510.001447476
0.0020173310.0017816560.001899493
0.0054905370.0065592470.006024892
0.0033519740.0017709820.002561478
0.0033519740.0025959810.002973977
0.0033519740.0042384350.003795204
0.0033519740.0036539180.003502946
0.0024972430.0035942580.003045751
0.0025855810.0033344440.002960012
0.0010609040.0004577120.000759308
0.0010609040.0023107830.001685843
0.0046139540.0059230670.00526851
0.0072090390.0039594810.00558426
0.0076682140.0079309940.007799604
0.0061988550.0060236980.006111277
0.0110200000.0091477220.010083861
0.0110201880.0095319340.010276061
0.0026042680.0034280910.00301618
0.0008646920.0008535370.000859114



where all those extraneous decimal places are due
to the normalization.

I would suggest that one consider the Sorenson rates
currently the "best" and the difference between theirs
and mine as an example of the "typical expected error".

Note the big difference in mine and Sorenson's for the
multiple markers like 464 and YCAII. This is because mine
actually take into account the difference in mutation
between the higher and lower allele values as actually
found in Ysearch for the different a, b, c, d, etc.

Doug

This thread: