We used genotypes from the two control cohorts in the WTCCC [2007] study to sample the UK Caucasian population without disease association. Data were available for 1,485 blood donors (the National Blood Service (NBS) sample) and 1,504 members of the 1958 British Birth Cohort. We considered the two samples, both separately and combined. We analyzed the autosomal data only, for which genotypes were available for 490,032 SNPs on the commercial release of the GeneChip 500K Affymetrix array. We excluded genotypes whose posterior probability according to the CHIAMO algorithm [WTCCC, 2007] was less than 0.95: this excluded 2.3% of genotypes in the NBS and 1% in the 58BC samples. In order to ensure that all test statistics had the χ2 distribution under the null hypothesis, we only used SNPs for which there were at least 10 subjects having each genotype, leaving 334,773 SNPs in the NBS, 335,331 in the 58BC and 359,491 in the combined samples.
