TABLE 7.
Power of the association test based on the statistic Tgene assuming three interacting disease loci within one gene under different sample sizes and different simulation schemes
| MS program
|
LD based (r2 = 0.9)
|
||||
|---|---|---|---|---|---|
| Disease model | (BL, GE) | n = 200 | n = 400 | n = 200 | n = 400 |
| Model 1 | (1, 0.7) | 0.465 (0.271) | 0.658 (0.685) | 0.866 (0.782) | 0.992 (0.857) |
| (1, 0.9) | 0.802 (0.481) | 0.925 (0.918) | 0.956 (0.867) | 1.000 (0.861) | |
| Model 2 | (1, 0.7) | 0.471 (0.299) | 0.629 (0.689) | 0.858 (0.774) | 0.994 (0.853) |
| (1, 0.9) | 0.808 (0.487) | 0.927 (0.876) | 0.960 (0.867) | 1.000 (0.876) | |
The power for the single SNP analysis is given in parentheses. BL, baseline effect; GE, genotypic effect. For the MS program, we set the first, fifth, and ninth SNPs as the three interacting loci. For the LD-based simulation, we simulate three LD blocks, one containing five SNPs, one containing three SNPs, and the other one containing two SNPs, with pairwise r2 = 0.9 for SNPs within each block. Three interacting disease loci are from each block with allele frequency PD = 0.3. The significance level is 0.05. Models 1 and 2 are defined in Equation 5.