The results in Figure 1 support the findings of a previous study that suggests that approximately 100 SNPs are required to resolve population structure in a sample [8]. However, these data also suggest that when SNPs with low MAFs are used, greater than 250 are required to obtain reliable results. We hypothesized that the SNPs with MAFs of near 0.50 may be much older than the origins of modern human races and therefore less informative than lower frequency SNPs that may have appeared closer to the time that modern Homo sapiens left Africa. However, these data suggest that on average high frequency SNPs provide better information. In fact, fewer SNPs of high frequency than low frequency are required to closely approach the STRP results. It appears that the potential greater sensitivity of more recent SNPs does not translate into an increase in useful information. It is also interesting to note that the similarity between the STRP and SNP runs never reaches 100%. The 96% similarity that is reached by these SNP sets is equivalent to an average difference in probability
