GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2005-11 > 1132933505
From: David Faux <>
Subject: Re: [DNA] autosomal testing leads to breakthrough
Date: Fri, 25 Nov 2005 07:45:05 -0800 (PST)
There is a tremendous interest in autosomal biogrographical testing. Although you may be "just experimenting" it certainly seems like an avenue worth exploring.
My statement about circa 100 STR or 400 SNP autosomal markers comes from extrapolations of simulation studies done with the program STRUCTURE on relatively unmixed populations. The data suggest that it would take about 400 carefully selected SNP markers (used in the currently available biogeographical test) or 100 highly selected STR markers (which give number of nucleotide pattern repeats not just a single value SNP) to provide a valid test. However with other approaches I don't know - all would need an algorithm that could convert the DNA findings to percentages of each geographic group (e.g., Sub - Saharan African). Thus I am unclear what comparing a person's scores to Burritt's tables would prove since the percentage differences by group is of such a low order of magnitude. Thus even if my marker was found in 7% of Koreans, 8% of Scandinavians, and 9% of Native Americans I don't think it would be justified to select the 9% just because it is marginally higher or be!
meets with my expectations.
Anyway, it still seems that at least this type of testing offers hope to many who have been disappointed by DNAPrint and ultimately you may find a mathematical way of predicting percentage for those who are unclear about their biogeographical ancestry. As it stands, I guess one needs to have genealogical clues before the data can be meaninfully interpreted.
I must again congratulate you on the rather ingenious tests you are bringing to the market. Cutting edge - and I suspect there is a lot more waiting in the wings.
Thomas Krahn <> wrote:
Thank you very much for your praise, but we only did the autosomal
testing itself. The interpretation was your work.
I wouldn't say that this was a new breakthrough, but it's simply
conventional population statistics based on the public forensic
Everybody can order an analysis of his autosomal STR markers from one of
the hundreds of paternity testing laboratories.
The resolution can only be as big as the degree of population
intermixture permits. Exactly there is the limitation of the autosomal
The most interesting finding of Mary Jo's autosomal STR genotype is a
significant influence of African ancestry next to the European main
I wouldn't get in detail with the resolution in distinct tribes. This
can do only Mary Jo herself with the background knowledge of her family
David Faux wrote:
> That is a very interesting finding. On this particular measure the data have to be interpreted in context. It would be very helpful if you could give us a listing of all your near matches so that we can better understand how the test works.
Mary Jo has used Brian Burritt's OmniPop Excel spreadsheet for
> It is generally assumed that only a test that explores 100 or more markers on 15 or more of the autosomes has a fighting chance of picking up low levels of minority ancestry.
Where did you get theese numbers from? In my experience the resolution
is mainly limited on the quality of the availlable databases. It depends
on the purity of non - mixed populations in the population dataset. More
markers will only help if you have absolutely monobreed populations in
your database, which is practically impossible because people always
mixed in any time of history. With the availlable databases you will
reach this limit after just a few markers.
The only way out of this limitation is narrowing the circle of ancestors
by using XSTR markers or with linked haploblocks.
>In theory it is highly unlikely that there would be enough autosomal DNA from an ancestor of 400 years ago to register on any test.
>However if there were multiple First Nations - European matings, or if like my family cousins were often the first choice for partners then the chances go up significantly.
Interesting approach. If we find separated minorities that don't mix
with other population groups, we could find seggregated autosomal
patterns. But this again is only valid, if we trust the paper trail. Or
do you see an independent method to proof this?
>In other words I am wondering whether you also had highly unlikely matches to say Australian Aborigines, or Koreans. We are always looking for breakthroughs, but I would need to know more before having the elder generations of my family tested to see if this !
> test can pick up our documented Canadian First Nations heritage.
Just to make this crystal clear to all readers: We're just
experimenting! There is absolutely no guarantee that the results are valid!
DNA-Fingerprint Tel.: +49-3329/697464
Fasanenstrasse 25a http://www.dna-fingerprint.com
14532 Stahnsdorf / Germany mailto: