2.7 KiB
2.7 KiB
1 | https://www.niehs.nih.gov/research/resources/software/biostatistics/trimmest/index.cfm | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
2 | TRIMMEST Table 1 | ||||||||||
3 | |||||||||||
4 | Carayol's Model | Number of SNPS in haplotype | Risk haplotype frequency | R1 used for generating simulated data | R2 used for generating simulated data | Patten of missingness* | Estimated risk haplotype frequency | Geometric mean of estimated R1 (95% CI) | Geometric mean of estimated R2 (95% CI) | Empirical coverage of nominal 95% confidence interval for R1 | Empirical coverage of nomincal 95% confidence interval for R2 |
5 | 1 | 3 | 0.4 | 1.5 | 4 | No missing | 0.4 | 1.51 (1.5, 1.52) | 4.02 (3.98, 4.07) | 0.94 | 0.944 |
6 | 5.3% missing | 0.4 | 1.51 (1.49, 1.52) | 4.01 (3.97, 4.06) | 0.954 | 0.944 | |||||
7 | 15% missing | 0.4 | 1.51 (1.5, 1.52) | 4.03 (3.98, 4.08) | 0.94 | 0.93 | |||||
8 | 2 | 3 | 0.1 | 4 | 11 | No missing | 0.1 | 4.03 (3.99, 4.06) | 11.14 (10.94, 11.34) | 0.946 | 0.94 |
9 | 5.3% missing | 0.1 | 4.03 (4, 4.07) | 11.15 (10.93, 11.36) | 0.956 | 0.942 | |||||
10 | 15% missing | 0.1 | 4.04 (4, 4.07) | 11.15 (10.93, 11.37) | 0.94 | 0.93 | |||||
11 | 3 | 6 | 0.4 | 1.5 | 4 | No missing | 0.4 | 1.5 (1.49, 1.51) | 4.01 (3.97, 4.06) | 0.968 | 0.954 |
12 | 2.7% missing | 0.4 | 1.5 (1.49, 1.51) | 4.01 (3.97, 4.06) | 0.962 | 0.958 | |||||
13 | 15% missing | 0.4 | 1.5 (1.49, 1.51) | 4.02 (3.97, 4.06) | 0.97 | 0.946 | |||||
14 | 4 | 6 | 0.1 | 4 | 11 | No missing | 0.1 | 4. (3.96, 4.04) | 10.9 (10.72, 11.09) | 0.936 | 0.944 |
15 | 2.7% missing | 0.1 | 3.99 (3.96, 4.03) | 10.87 (10.69, 11.07) | 0.942 | 0.94 | |||||
16 | 15% missing | 0.1 | 4.01 (3.97, 4.05) | 10.96 (10.76, 11.16) | 0.936 | 0.948 | |||||
17 | *This column represents the percentage of genotypes missing randomly, which is different from the missingness defined by Carayol et al. (the probability for an individual to have at least one unobserved SNP in a haplotype). Our scenarios of 5.3% and 15% missing genotypes correspond to their 15% and 39% missingness respectively. | ||||||||||
18 | |||||||||||
19 | |||||||||||
20 | |||||||||||
21 | TRIMMEST Table 2 | ||||||||||
22 | Pattern of missingness* | Disequilibrium coefficient | Frequence of genotypes with 0, 1, and 2 copies of the risk haplotypes | Geometric mean of estimated R1 (95% CI) | Geometric mean of estimated R2 (95% CI) | Empirical coverage of nominal 95% confidence interval for (R1, R2) | |||||
23 | No missing | 0.1 | (0.46, 0.28, 0.26) | 1.50(1.49, 1.51) | 4.02(3.97, 4.06) | 0.952 | |||||
24 | 20% missing | (0.46, 0.28, 0.26) | 1.51 (1.50, 1.52) | 4.03 (3.98, 4.08) | 0.959 | ||||||
25 | No missing | -0.1 | (0.26, 0.68, 0.06) | 1.550 (1.450, 1.51) | 4.01(3.98, 4.04) | 0.948 | |||||
26 | 20% missing | (0.26, 0.68, 0.06) | 1.51(1.50, 1.52) | 4.01(3.98,4.04) | 0.946 | ||||||
27 | Note: Simulation conditions is Model 1 of Carayol et al. A total of 6 haplotypes comprised of 3 SNPs; risk haplotype frequency=0.4 * Our 20% missing genotypes correspond to Carayol et al's 49% missingness. | ||||||||||
28 |