False positive rates for rare variant association studies when variants are identified only in cases or in both cases and controls for variants with a mixture of frequencies.

aThe number of controls is equal to the number of cases.bNumber of rare variants with frequency ��1% observed per haplotype pool averaged over 100 haplotype pools.cTotal frequency of rare variants with frequency ��1% observed per haplotype pool averaged over 100 haplotype pools.Coalescent simulations with scaled mutation rates �� ranging between 4 and 12 were used to generate genotype data for rare variants with frequencies between 0.0001 and 0.01 for samples of N = 100, 200, 500, and 1,000 cases. The false positive rates are displayed when variant discovery is carried out in only cases via sequencing and the discovered variants are genotyped in controls (upper panel) and when both cases and controls are sequenced to discover rare variants (lower panel). Analyses were carried out using the Cochran�CArmitage test for trend (see Methods). The false positive rates were evaluated for an �� = 0.05 and based upon 100,000 replicates.