Pass one has three reference populations. Asian, African and European. Lets imagine a person who is about admixed 33.3% each. The next pass would take the DNA flagged as Asian and refine the determination by using however many reasonably distinct populations were available. This pass would then do the same for the persons African and European portions.
One of the problems I've read about is that things get squirrely if you include too many populations which is likely why companies like Ancestry, 23andMe and FTDNA try to limit the number of populations they use. By doing the admixture testing in multiple passes they can have more populations in each grouping without shorting any of the others.
It makes sense to me, but I don't really know what I'm talking about.

Leave a comment: