DNAprint Genomics (DNAG)

[mingwan0]

Frog, I think you are as aware of the publicly available validation details as I am. The results of the Validation Studies undertaken in conjunction with the National Center for Forensic Science and the San Diego Police Department Crime Lab on version 2.0 are available on Roger's website:

http://archiver.rootsweb.com/th/read/GENEALOGY-DNA/2004-01/1073420735

Incidentally, Experiment II (Comparing Self-Reported Race with the Major Admix Percentage) tells us a little more detail about the parental groups used.

I am not aware of any such information for 2.5 other than the information that was posted on the ABD website and RootsWeb.

As you might expect, I take issue with your assertion that there were "significant accuracy issues with the first version." The only genuine flaw has been discussed by Tony Frudakis:

http://archiver.rootsweb.com/th/read/GENEALOGY-DNA/2004-01/1073420735

"Several months ago we realized that the test was missing African for a small minority of individuals in the teen-30 percent range due to a flaw in the algorithm used. This flaw affected only African admixture and only for a small fraction of individuals with Native American AND African ancestry AND of relatively high (Native American + African) admixture (i.e. very high heterozygosity). It did not affect individuals with substantial African admixture, whether they fit the other criteria just mentioned or not. We were originally alterted to this problem from 2 individuals for whom we confirmed there was an African error. Since then, there have been 2 or 3 others that have stepped forward and the total (of about 6,000 tests run so far) is about 5."

"The flaw, in retrospect, was obvious and very subtle. It was not flagged before because it affects such a very, very small number of possible multilocus genotypes (i.e. people). Since the bug that caused this problem was easily identified, it was easily fixed several months ago - and I think most of you are aware an improvement was made several months ago - the "new" algorithm is what resulted, and nothing has changed since except the style of the presentation.. We havent had any problems detecting dilute African ancestry in these rare individuals of substantial Native American/European admixture since (or anyone else)."

In terms of accuracy and precision of ABD, perhaps you need to refamiliarize yourself with the relevant section of the ABD website:

http://www.ancestrybydna.com/simulation.asp

I believe that the validations do have intrinsic scientific value, as does the test itself.