Identifying the genetic basis of human adaptation has remained a central focal point of modern population genetics. One major area of interest has been the use of polymorphism data to detect so-called "footprints" of selective sweeps - patterns produced as a beneficial mutation arises and rapidly fixes in the population. Based on numerous simulation studies and power analyses, the necessary sample size for achieving appreciable power has been shown to vary from a few individuals to a few dozen, depending on the test statistic. And yet, the sequencing of multiple copies of a single region, or of multiple genomes as is now often the case, incurs considerable cost. Enard et al. (2010) have recently proposed a method to identify patterns of selective sweeps using a single genome - and apply this approach to human and non-human primates (chimpanzee, orangutan, and macaque). They employ essentially a modification of the Hudson, Kreitman, and Aguade test - using heterozygous single nucleotide polymorphisms from single individuals, and divergence data from two closely related species (human-chimpanzee, human-orangutan, and human-macaque). Given the potential importance of this finding, we here investigate the properties of this statistic. We demonstrate through simulation that this approach is neither robust to demography nor background selection; nor is it robust to variable recombination rates.
Keywords: adaptation; demography; selective sweeps; statistical inference.