GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-03 > 1173898826


From: "Alfred A. Aburto Jr." <>
Subject: Re: [DNA] OmniPop caution received from FamilyTreeDNA
Date: Wed, 14 Mar 2007 12:00:01 -0800
References: <cdb.9c8bb2a.33283343@aol.com><45F6F8C5.9030209@sbcglobal.net> <45F767C0.9010501@mchsi.com><45F78B4D.7020001@sbcglobal.net><40A3F6B4-5ED4-4C59-BF0B-E33FC2572CEC@vizachero.com>
In-Reply-To: <40A3F6B4-5ED4-4C59-BF0B-E33FC2572CEC@vizachero.com>


> Vincent Vizachero wrote:

>On Mar 14, 2007, at 12:42 AM, Alfred A. Aburto Jr. wrote:
>
>
>
>>I think so too. OmniPop and the RCMP employ established databases
>>published in the technical literature and I'm sure the analysis
>>conducted (the results) are correctly done.
>>
>>
>
>Right. The problem isn't with OmniPop or its data: it is a quite
>good tool for its conceived purpose. The problem is that people are
>using it to do something for which it is not suitable: admixture
>estimation.
>
>

Yes, OmniPop, and the RCMP Profiler website, give probabilities for the
populations included nothing more, except as Ann indicated one can
compare frequencies of various markers for various populations (RCMP
Profiler website). They do not give one the percent for which one
belongs to or matches a given population group because the number of
people in the population group affects the result (thanks to John
Chandler for pointing this out to me) . For a given probability a very
large population group may give a relatively large percentage while a
small population group for that same probability will give a
correspondingly smaller percentage. With estimates of population sizes
OmniPop and RCMP's Profiler could give admixture estimates, but
unfortunately they do not provide this additional information
(population sizes for their data), and we must be careful in how we
interpret the results. The probabilities are still useful in themselves,
but one must be careful not to think if they have a high probability
result for a given population that they actually derive from that
population (they may or may not would be the correct answer, depending
on the actual estimated population sizes, which are not given in these
programs).
Al


>Vince
>
>

This thread: