Table II.
Log-linear model assumption |
HWD coefficient (D11) |
Simulated R1 |
Pattern of missingness1 |
Mean total number of transmission tetratypes |
Mean number of triads with multiple transmission tetratypes |
Power (1df) |
Empirical coverage of nominal 95% confidence interval |
Frequency of carriers of the risk haplotype2 |
Geometric mean of estimated R1 (95% CI)3 |
||
---|---|---|---|---|---|---|---|---|---|---|---|
HWE | -0.1 | 1 | None | 419.4 | 19.1 | 0.071 | 0.929 | 0.36 | 0.48 | 0.16 | 1.00(0.99,1.00) |
20% g | 917.4 | 205.0 | 0.073 | 0.927 | 0.36 | 0.48 | 0.16 | 1.00 (1.00,1.00) | |||
-0.1 | 1.414 | None | 419.3 | 19.0 | 0.988 | 0.876 | 0.39 | 0.47 | 0.14 | 1.51(1.5,1.51) | |
20% g | 921.1 | 205.0 | 0.983 | 0.873 | 0.39 | 0.47 | 0.14 | 1.51(1.5,1.51) | |||
0.1 | 1 | None | 407.7 | 7.6 | 0.021 | 0.979 | 0.36 | 0.48 | 0.16 | 1.00 (1.00,1.01) | |
20% g | 863.1 | 191.9 | 0.021 | 0.979 | 0.36 | 0.48 | 0.16 | 1.00 (1.00,1.00) | |||
0.1 | 1.414 | None | 407.1 | 7.0 | 0.618 | 0.916 | 0.33 | 0.49 | 0.18 | 1.29(1.29,1.29) | |
20% g | 854.0 | 189.7 | 0.570 | 0.912 | 0.32 | 0.49 | 0.19 | 1.28(1.28,1.29) | |||
Random Mating |
-0.1 | 1 | None | same as above | 0.049 | 0.951 | 0.26 | 0.68 | 0.06 | 1.00 (1.00,1.00) | |
20% g | 0.053 | 0.947 | 0.26 | 0.68 | 0.06 | 1.00 (1.00,1.00) | |||||
-0.1 | 1.414 | None | 0.982 | 0.946 | 0.26 | 0.68 | 0.06 | 1.41(1.41,1.42) | |||
20% g | 0.977 | 0.949 | 0.26 | 0.68 | 0.06 | 1.41(1.41,1.42) | |||||
0.1 | 1 | None | 0.050 | 0.950 | 0.46 | 0.28 | 0.26 | 1.00 (1.00,1.01) | |||
20% g | 0.048 | 0.952 | 0.46 | 0.28 | 0.26 | 1.00 (1.00,1.01) | |||||
0.1 | 1.414 | None | 0.744 | 0.952 | 0.46 | 0.28 | 0.26 | 1.42(1.41,1.42) | |||
20% g | 0.702 | 0.953 | 0.46 | 0.28 | 0.26 | 1.42(1.41,1.42) |
None: no missing data; 20% g: 20% of SNP genotypes are missing randomly.
Expected frequencies of diplotypes with 0, 1, 2 copies of the risk haplotype are 0.26, 0.68, and 0.06, respectively, when D11 =-0.1 and 0.46, 0.28, and 0.26, respectively, when D11 = 0.1.
The 95% CI is based on the empirical standard error calculated using the estimates from 5000 independent simulation.