False positive rates for association studies when variant identification is carried out in only cases or in both cases and controls for gene(s) with a fixed number of neutral variants.

Results are shown for gene(s) with M = 10, 20, and 30 neutral variants with equal population frequencies of 0.001, 0.002, and 0.005, for N = 100, 200, 500, and 1,000 cases, and an equal number of controls. The assumption is made that the variants reside on separate haplotypes. The upper panel shows the false positive rates when only cases are used for variant discovery and the discovered variants are genotyped in controls. The lower panel shows the false positive rates when both case and controls are sequenced to discover rare variants. Analyses were carried out using the Cochran–Armitage test for trend (see Methods). The false positive rates were evaluated for an α = 0.05 and based upon 100,000 replicates.