Fig. 3
Comparison of the asymptotic power calculations with the power attained by Haplin and EMIM in data simulations. The power was calculated for different child-parent configurations, assuming a MAF of 0.2 and a nominal significance level of 0.05. The results were based on 500 case families and, when applicable, 500 unrelated control families. All simulations were based on 10,000 replicates of data for a single SNP. Asymp: Power calculations in Haplin, based on asymptotic approximations (Haplin function hapPowerAsymp); Haplin: Power calculations in Haplin, based on data simulations. The power is the proportion of tests rejected by Haplin (Haplin functions hapRun and hapPower); EMIM: Power calculations based on data simulations in Haplin (Haplin function hapSim). The power is the proportion of tests rejected by EMIM. a Child effects (RR >1); b Child effects, adjusting for maternal effects (RR >1 and RR(M)=1); c PoO effects (RRM/RRF>1 and RRF=1); d PoO effects, adjusting for maternal effects (RRM/RRF>1 and RRF=RR(M)=1); e Maternal effects, adjusting for child effects (RR(M)>1 and RR = 1); f Maternal effects, adjusting for PoO effects (RR(M)>1 and RRM=RRF=1). The power was calculated at relative risks/relative risk ratios of 1,1.1,1.2,…,2. Intermediate values correspond to line segments joining two adjacent points. Note that for all study designs, the power was calculated based on asymptotic approximations in Haplin, as well as simulations where both Haplin and EMIM were used to analyze the genetic data. The lines for Asymp, Haplin and EMIM are nearly overlapping, demonstrating consistent results