Common SNPs (MAF>25%) exhibited significantly fewer false positives than expected by chance at all thresholds tested (e.g., 10-4, 10-5, and 10-6; Figure 1). Rare SNPs exhibited more variability in the number of false positives than common SNPs, but did not have significantly more of false-positive results than expected by chance. Interestingly, when SNPs with 5% missing data were analyzed, the findings did not differ from the results using SNPs with almost no missing data (data not shown). Given the marked variability in the 1% MAF, the relationship between numbers of individuals homozygous for the 1% allele and number of false positives was examined, but no trend was present.
