GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2007-03 > 1173830117
Subject: Re: [DNA] OmniPop caution received from FamilyTreeDNA
Date: Tue, 13 Mar 2007 19:55:17 EDT
In a message dated 3/13/2007 10:03:56 AM Pacific Daylight Time,
writes (quoting Bennett Greenspan):
> "We are offering the Autosomal markers because we have been asked for them,
> to match against some CODIS markers I presume, but we are not very impressed
> with the database that OMNIPOP has and we feel that it can provide
> information so we don’t have it interactive on our site...
The trouble is not so much with the database, which is very valuable WHEN
USED FOR THE PURPOSE INTENDED, looking up the frequency of various alleles in
different populations. The problem is with the way people over-interpret the
results. Brian Burritt, the creator of OmniPop, has participated in some threads
on DNA-Forums. Here are a couple of quotes from him:
"As far as I know, neither [ENSFI or OmniPop] is used to do any ethnicity
testing for criminal cases. I work for a police department and use OmniPop to
double check calculation made with our official software."
"So, let's see. I create a unique database of population data as a population
genetic research and forensic tool to be used by whoever finds it useful. The
genetic genealogy community catches wind of it (DNA Tribes used as the core
and inspiration for their company). Then, I get made fun of and insulted
because it's not the perfect genetic genealogy tool. Once again, and listen closely,
when have I ever claimed that it was? These faulty conclusions and shaky
assumptions are your own. It has its pros and cons like every other test out
there. If it's useful for you, great. If not, don't use it."
The CODIS markers were not selected to be "Ancestry Informative Markers" --
they were chosen because they can, in combination, uniquely identify an
individual. This is because every population has a wide spread of alleles. They can
also be used to verify close relationships: parent/child and siblings. For
example, paternity index calculations need to account for the frequency of alleles
in the relevant population: if you have a 15 for D3S1358 and your alleged
father also has a 15, how often would that occur just by chance if you picked a
random man off the street?
Technical articles using STRs to infer ancestry use hundreds of markers --
and they get the best results when they pick markers that show big differences
in various populations.
AOL now offers free email to everyone.
Find out more about what's free from AOL at http://www.aol.com.