Figure 4From: Trick or treat: The effect of placebo on the power of pharmacogenetic association studiesThe effect of the number ( M ) of markers typed (three or five) on the power of detecting an association between the markers and the response to the drug. The results are based on 100 simulations. The estimated power is shown for M = 3 (-◆-), M = 5 (-■-) and the causative single nucleotide polymorphism (-x-). The shaded regions denote designs for which the difference between M = 3 and M = 5 is significant. (a) Random selection, genotype relative risk (GRR) = 3, f2 = 0.6, f0 = 0.2; (b) Categories strategy, GRR = 3, f2 = 0.6, f0 = 0.2; (c) Random selection, GRR = 3, f2 = 0.8, f0 = 0.266; (d) Categories strategy, GRR = 3, f2 = 0.8, f0 = 0.266; (e) Random selection, GRR = 4, f2 = 0.8, f0 = 0.2; (f) Categories strategy, GRR = 4, f2 = 0.8, f0 = 0.2.Back to article page