I said that, and it's true. There is no one-to-one mapping between those genes and a person's race. Thus, no definitive genetic test for ethnicity exists. And even if you could have such a thing, then when we start using it as a precondition of receiving, say, affirmative action benefits, then and only then will the author have something resembling a cogent case.
We don't make black people "prove" that they are black in order to receive special benefits from society, and therefore this notion that we should make gay people "prove" that they are gay falls flat on its face.
It's a shame, because you can make a very good case for not singling out groups for preferences - this article isn't it.
That's because the argument has to apply to all groups, and the aspinal double-orchidectomy crowd dares not present the case for simply getting the government out of the whole mess.