Gene Expression: Race: the current consensusII. Clusters and race
The fact that humans cluster together based on genetic information could, in theory, be entirely orthoganal to the concept of race. However, at least in the United State (where this has been explicitly tested), this is not the case. The most important reason for this, in my mind, is that the ancestors of European-Americans and African-Americans were not randomly sampled from the globe (there's a bias towards points on the globe that are quite distant), and this non-random sampling accentuates the genetic differences between the two groups. But in any case, the reasons for this are irrelevant to the argument; let's look at the data.
The basis for this assertion comes from a paper (open access) by a different set of researchers at Stanford, who assembled a group of Americans who identified themselves as either African-American, white, East Asian, or Hispanic. They followed a similar protocal as the studies in the first section-- they took DNA from all individuals, looked a hundreds of different DNA variants, and applied a clustering algorithm. They then looked to see if their clusters corresponded to self-reported group. And indeed, in 3631 out of 3636 cases (99.85%), the individuals were clustered by the algorithm into the "correct" racial group.
I read an interview with one of the authors of the paper and he said that there were more gender discrepancies than racial ones.