Table 1.
Type I Error Rates of Unadjusted and Robust Versions of Burden, SKAT, and SKAT-O and Hybrid Method
Case: Control | Burden | Robust Burden | SKAT | Robust SKAT | SKAT-O | Robust SKAT-O | Hybrid SKAT-O |
---|---|---|---|---|---|---|---|
1:1 | 1.00 | 1.00 | 0.99 | 0.99 | 1.11 | 1.11 | 1.09 |
1:9 | 0.99 | 1.00 | 1.01 | 1.01 | 1.13 | 1.13 | 1.09 |
1:49 | 1.02 | 0.95 | 1.44 | 1.22 | 1.44 | 1.23 | 1.27 |
1:99 | 1.07 | 0.91 | 1.92 | 1.41 | 1.82 | 1.33 | 1.53 |
1:1 | 1.02 | 1.00 | 0.99 | 1.03 | 1.27 | 1.32 | 1.27 |
1:9 | 1.12 | 0.99 | 1.39 | 1.14 | 1.65 | 1.40 | 1.52 |
1:49 | 2.43 | 0.97 | 6.31 | 1.65 | 6.16 | 1.79 | 4.54 |
1:99 | 3.95 | 1.02 | 13.48 | 2.13 | 12.77 | 2.17 | 8.89 |
1:1 | 1.11 | 1.03 | 1.24 | 1.54 | 1.38 | 1.38 | 1.40 |
1:9 | 1.29 | 0.77 | 2.47 | 1.45 | 2.51 | 1.49 | 2.23 |
1:49 | 6.88 | 1.06 | 28.27 | 1.91 | 23.70 | 1.98 | 16.69 |
1:99 | 16.34 | 0.90 | 89.53 | 1.81 | 71.32 | 1.60 | 42.59 |
A total of 107 datasets of 1 kb regions were generated to estimate type I error rates. Each cell represents an empirical type I error rate divided by significance level . The sample size was 50,000.