GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2003-09 > 1062515477


From: Charles <>
Subject: Re: [DNA] An idea! Why not reprocess the genotype data with the new algorithm for each prior customer's test data instead of a complete new retest?
Date: Tue, 02 Sep 2003 11:13:59 -0400
References: <410-220039229276421@earthlink.net>


All:

I think everyone agrees that the actual Genotype data table that
DNAPrint reports to each person is factually accurate.

What has been questioned here at various times is what interpretation
DNAPrint's proprietary algorithm is placing on that data as to what that
Genotype data means in assaying a person's genome into the 4 population
groups reported by DNAPrint.

I can imagine that tweaking changes in this type test could come in two
or three areas. 1. Adding or removing certain markers from the test. 2.
Modifying the frequency of occurence ratios for the various genotype
alleles for the various markers based on new people samples or research
information. 3. Changes or "bug fixes" could also be made in the
computer program which mathematically processes the data.

Changing which markers were used and/or the allele frequencies for the
various population groups for those markers could affect the outcome
dramatically. If not many changes have been made in the number or which
markers which were used at various times over the last 2 years, but
significant changes were made in the occurence freqency ratios for the
various alleles at the various markers, then a reprocessing of the
earlier genotype data through the latest computer model would be
possible, in my opinion.

As an idea for DNAPrint to consider, I would suggest that the Genotype
data obtained from earlier versions of the 2.0 test be re-run at no
charge to the customer through any changed computer model in any new
allele frequency values matrix and the changed computer program
algorithm to get a new set of test results for us customers. Recomputing
would be a lot less costly to DNAPrint then re-testing. If a few markers
were changed along the way, I guess they could run the new algorithm
using the earlier Genotype data with "FL/Failed" or "NT/Not Tested" in
the locations for any new marker the person doesn't have results for,
unless of course these new markers added are very dominant and
significant to the overall assay results for certain population groups.
But I was always told by DNAPrint folks that changes for only a few
markers did not affect the overall test significantly.

Therefore if test versions 2.01, 2.02, 2.03, 2.1, 2.2, 2.3, etc., exist
behind the scenes at DNAPrint but have not been divulged by DNAPrint ...
but have been used over the past year or so, then DNAPrint could re-run
a person's "genotype data table" through the newest version ... say 2.4
or 2.5 ... and get the latest results without completely retesting
everyone. This is all assuming that significant numbers of markers have
not been changed in any new version of the test and that just the
frequency of occurence ratio allele values for the various alleles at
the various markers have been changed because of new sampling data or
research, etc.

Just an idea I am throwing out for consideration for comment by those
more familiar with the science used and this test.

Charles


ernest hurst wrote:
>
> Some mighty good detective work there, Cecelia. I'll be real interested to hear them explain this away. Looks to me like all of us "early tested" folks should get at free retest at current standards, or better yet get money refunded since this is probably not a viable test, in any version.
>
> Ernie
>
> ==============================
> To join Ancestry.com and access our 1.2 billion online genealogy records, go to:
> http://www.ancestry.com/rd/redir.asp?targetid=571&sourceid=1237

This thread: