The pooling approach used herein provides evidence for good assessment of allele frequency differences and variation in these estimates. SNP allele frequency assessments made herein display modest variability and good fits between individual and pooled genotyping with mean correlation of 0.98+/−0.002 (standard error, SEM, Fig. 1). Validating studies with similar arrays add to confidence in this data [13], [14], [26]–[34]. SEM for the variation among three replicate studies of each DNA pool was +/−0.03. SEM for the variation between the ca. 20 pools studied for each ethnicity/phenotype group was +/−0.02. These estimates of variability allowed us to estimate 0.8 and 0.9 power to detect 5 and 10% allele frequency differences in the African American sample sizes described here. We had 0.76 and 0.99 power to detect 5 and 10% allele frequency differences in European American samples. Corresponding false negative probabilities for approach 1 (converge then cluster) are thus 0.39 and 0.11, since this approach requires nominally positive results for the same SNP from both samples. Statistical power for the analysis of these samples can also be calculated using “gene detective”
