No evidence of selection in African Americans since their diversion from African peoples. This study contradicts an earlier one, using 15 times as many subjects (N=29k !) and properly correcting for multiple hypotheses. Note that the previous study had corrected for 20 hypotheses (that is, SNPs with possibly enhanced frequency in African Americans), rather than the appropriate 5000. Since the p-value (or whatever test statistic is used) must be multiplied by the number of hypotheses to find the true probability that the divergence is not due to chance, this is a correction by a factor of 250! Not the sort of thing that one should be messing up.
Furthermore, this is a useful corrective to the idea that African Americans were naturally selected for certain traits as fieldworkers that lead to phenotypes present today (e.g. more fast twitch muscle fibers, greater propensity for hypertension, etc).