Blank slatist genomics dilemma: if PGS algos are trained only on Euros, they work poorly for Africans, but also can't prove IQ gaps' genetic origin. Once PGSs work equally across groups, one cannot hide behind LD decay for the population PGS comparisons.https://www.technologyreview.com/s/612322/white-people-only-dna-tests-show-how-unequal-science-has-become/ …
-
i'm not sure they can honestly hide behind LD decay right now. if LD decay is the main driver of differences between racial groups why does the direction of bias result in average PGS that gets rank order of racial groups correct?
That's the big question for them to answer. So far, it's been correct every time to my knowledge including with each iteration of the GWASs. Education/IQ, height, schizophrenia. Any others? I think there are a few more.
