DNAprint Genomics (DNAG)

[mingwan0]

I know that somebody posted this link recently, but it is worth repeating it as the comments by Tony Frudakis are worth (re)reading:

http://dienekes.blogspot.com/2004/08/euro-dna-test.html

Here are some (large) excerpts of Tony's posts (the others are interesting as well):

PART I – I dont mean to infiltrate your group here - I want you and every one else to discuss the test objectively as you seem to do, without interference from us here at DNAPrint - but I thought I could add something here just this one time. In reference to the choice of population model - we did in fact use Pritchard's STRUCTURE on a pan-European collection of samples to define the most appropriate population model for the bulk of Europeans. With k=4, our results were very similar to Rosenbergs, except we did not include the population isolates for obvious reasons (i.e. they are isolates, and therefore by definition not representative of the bulk of Europeans). By the way, we also used Principle Components Analysis like Cavalli-Sforza did, (reviewd in Joblings 2004 textbook), and got the same result - most of the variation is in a northwest to southeast orientation. (cont'd part II)

PART II - STRUCTURE broke the samples out into this same orientation, as shown on the website for the product, and the NOR and SA groups at the polar extremes of this range are associated with anthropometric traits (iris color) (also shown on the website). I think that speaks for itself. On top of this, the predominant archaeological clines are in the northwest to southeast orientation, as are classical blood group and Y-chromosome haplogroup distributions. Clearly, most of the genetic variation is distributed throughout Europe in this orientation. Maybe in the future we can make the test more complex by adding more populations and more markers to capture the second and third principle components of variation. We actually have this data and these markers - but it is not a practical test yet. Larger, more complex population models require more markers however, so it would be a much more expensive test. (cont’d part III).

Part III - You have to start somewhere so we started with the the principle component
As for excluded groups - not only did we not include the Basques, butwe also did not specifically include the Sardinians, nor the Roma, though the spread of the latter throughout Europe likely impact the distribution of ancestry we have so far observed (such as the high percentage in Iberia, where the Roma were known to settle en masse). We are actually testing how the Basques type as of this week, but Im afraid nobody could possibly know what to expect here. Basque and Sardininan ancestry must be the 9th and 10th or even greater principle components of variation in Europe (Im being facetious, but the point Im making is clear). Due to their supposed paleolith connections, should they type of more "NOR" or would STRUCTURE kick them out into their own group with our markers as it did for Rosenberg? Is the EURO-1.0 population model completely inappropriate for them? (contd part IV)

Part IV - With the chosen population model of EURO 1.0, they will type based on their genetic distance to each of the 4 groups. If the genetic distance to the closest group is still very large, we can say the model is inappropriate for this population. Similarly, it would also be inappropriate for a 100% sub-Saharan African, who could insist on taking the test and would be shoe-horned into the 4 EUropean groups even though they are not European and have no European admixture. For this reason, since EURO 1.0 is a hierarchical test requiring a prior fit with the population model, we do our best to pre-screen customers for this fit. This is why Ancestrybydna 2.5 is required. Unfortunately for Basques, this screen is impossible, and their ancestry would be shoe-horned into the NOR, MED, MIDEAS, SA model. However, unlike sub-Saharan Africans, the genetic distance between Basqueness and NOR/MED is almost certainly (cont’d part V)

Part V very low, and the test would therefore be reporting a more ancient connection than for most other EUropean customers. After all, EEuropean paleoliths and neoliths once shared a common non-African ancestor, did they not?

Icelanders are another isolate, with Scottish and Scandinavian roots, who type mainly with NOR ancestry with EURO-DNA 1.0. In this case, we know this is the expected result. The data for the Basques and Icelanders should be on the website soon.

Cheers and thanks for thinking about our science. Continue to critisize, evaluate, and think - good job to you all. I believe this is the most sophisticated discussion of our technology available other than from us and in the scientific literature.
Tony Frudakis

Scoob - that is an interesting analysis. Ill pass that on to Paul McKeigue and Mark Shriver and see what they think - they are two of the premier thought leaders in the field of molecular anthropology, and our close collaborators. Ill come back with their reactions if you like.

Part III - Hilde – I wonder if the Galicia were recent offshoots of more northerly populations compared to other neighboring ethnicities? This is the type of question I believe our test is well suited for. But as Dianekes suggests, there is plenty to think about when interpreting the test results. I cannot monitor the discussion daily, so I will trust that if there is a question of significance related to the test, or some confusion on the theory, perhaps Dianekes can email me so that I can provide my input. From my experience here, it is clear to me I need to reword part of the website so that our method of developing the test is more clear. As soon as we get the data on the Basques and Icelanders, we will post them and Id be happy to interpret with you folks if emailed. I just wanted to stop in and provide some input these first days the test is out.

Some interesting insights. The comment about testing the Basques "this week" was dated 31st August so it is very current.