Figure 1.
Population structures of the different scenarios.
Samples are represented on the first two principal components (PCs) estimated on the genotype data.
Figure 2.
Scenario 1 (One homogeneous population).
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Powers of the methods in function of the increasing relative risk.
Figure 3.
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Powers of the methods in function of the increasing relative risk.
Figure 4.
Scenario 3 (Two fairly distant discrete populations).
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Powers of the methods in function of the increasing relative risk.
Figure 5.
Scenario 4 (Two very distant discrete populations).
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Powers of the methods in function of the increasing relative risk.
Figure 6.
Scenario 5 (Hierarchical structure).
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Powers of the methods in function of the increasing relative risk.
Figure 7.
Scenario 6 (Varying proportions of cases/controls).
A - False positive rates of the methods. The plain black line represents the 5% level at which the tests were conducted. The dashed black lines are the 95% confidence intervals for this level. B - Evolution of the power (with RR = 2) of the methods in function of the proportion of cases in pop1. Note that all the Regression methods being equivalent for this scenario, we summarize the results for these methods under the name ‘Reg’ only.
Table 1.
Summary table.