Table 1.
Significance threshold α′ (×10−5) and effective number of tests (×103) based on Bon-ferroni correction (ENTB). 95% confidence intervals are given in parentheses. Bold figures are the proportion of Effective/Actual number of tests. Figures are based on 50,000 re-samples of the disease indicator under the null hypothesis.
| Dataset | Sample size #Cases/controls |
Overall error rate α = |
||||
|---|---|---|---|---|---|---|
| 1% | 5% | |||||
| Chinese population High resolution 5 × 10−4 ppm 16,118 variables 836 spectra |
50/50 | α′ | 0.48 (0.45;0.52) | 13% | 2.12 (2.03;2.19) | 15% |
| ENTB | 2.08 (1.92;2.24) | 2.36 (2.28;2.46) | ||||
| 100/100 | α′ | 0.35 (0.33;0.39) | 17% | 1.64 (1.58;1.70) | 19% | |
| ENTB | 2.82 (2.58;3.05) | 3.06 (2.95;3.16) | ||||
| 200/200 | α′ | 0.25 (0.23;0.27) | 25% | 1.24 (1.19;1.31) | 25% | |
| ENTB | 3.96 (3.69;4.26) | 4.02 (3.83;4.19) | ||||
| Chinese population Medium resolution 10−3 ppm 7,100 variables 836 spectra |
50/50 | α′ | 0.87 (0.82;0.94) | 16% | 3.81 (3.68;3.92) | 18% |
| ENTB | 1.14 (1.06;1.23) | 1.31 (1.27;1.36) | ||||
| 100/100 | α′ | 0.66 (0.61;0.72) | 21% | 3.22 (3.09;3.36) | 22% | |
| ENTB | 1.51 (1.40;1.64) | 1.55 (1.49;1.62) | ||||
| 200/200 | α′ | 0.50 (0.46;0.54) | 28% | 2.48 (2.37;2.58) | 28% | |
| ENTB | 2.00 (1.84;2.16) | 2.02 (1.93;2.11) | ||||
| US population Medium resolution 10−3 ppm 7,100 variables 2164 spectra |
50/50 | α′ | 1.04 (0.95;1.14) | 14% | 4.43 (4.28;4.59) | 16% |
| ENTB | 0.97 (0.88;1.05) | 1.13 (1.09;1.17) | ||||
| 100/100 | α′ | 0.81 (0.73;0.86) | 17% | 3.61 (3.46;3.74) | 20% | |
| ENTB | 1.24 (1.16;1.37) | 1.39 (1.34;1.44) | ||||
| 200/200 | α′ | 0.65 (0.60;0.71) | 22% | 2.98 (2.84;3.07) | 24% | |
| ENTB | 1.54 (1.41;1.68) | 1.68 (1.63;1.75) | ||||
| 500/500 | α′ | 0.46 (0.44;0.51) | 30% | 2.41 (2.31;2.51) | 29% | |
| ENTB | 2.15 (1.95;2.30) | 2.07 (1.99;2.16) | ||||