I agree with you, but I would also note that the original person being replied to seems to be committing a common colloquial misunderstanding in regards to social constructions; just because something is socially constructed does not mean it's not real. Race is absolutely real, regardless of the degree to which genetics does or doesn't correlate with it, because we make it real in our day to day interactions.

But your point is succinct: just because you can group people using genetic markers into clusters that align well with our society's typical social categorization of race does by no means validate a genetic basis for all the other sociocultural baggage that we associate with those categories. The most it seems to do particularly in regards to the social category of race is show that genetic markers useful for identifying kinship (I'd file that under "duh"). It's interesting to note, for me, that the analysis of the first paper works just fine for 2 clusters as well (basically East Asian and everyone else), but those super-racial groupings don't exist in the US to my knowledge.