Table 3.
Per-metabolite statistical power, calculated over the 50 replications of the disease model, for the single (Table 3-a) and the multi-metabolite models (Table 3-b). False positive (FP) rate is defined as the mean number of false positive associations as a proportion of the number of significant variables. Population size did not allow investigation of sample sizes >50 at a prevalence of 10%. FWER is set to 1%.
Table 3-a One disease associated metabolite (hippurate).
| Prevalence | Sample Size | 10% |
30% |
50% |
|||
|---|---|---|---|---|---|---|---|
| Power | FP Rate | Power | FP Rate | Power | FP Rate | ||
| O2PLS Bootstrap | 50/50 | 100% | 33.4% | 100% | 29.1% | 100% | 30.1% |
| 100/100 | N.A. | N.A. | 100% | 36.7% | 100% | 53.1% | |
| 200/200 | N.A. | N.A. | 100% | 50.9% | 100% | 69.8% | |
| O2PLS Permutation | 50/50 | 98% | 9.3% | 98% | 9.2% | 100% | 3.5% |
| 100/100 | N.A. | N.A. | 100% | 6.1% | 100% | 4.7% | |
| 200/200 | N.A. | N.A. | 100% | 4.9% | 100% | 6.0% | |
| T-test Bonferroni | 50/50 | 68% | 0.4% | 66% | 0.3% | 94% | 0.4% |
| 100/100 | N.A. | N.A. | 100% | 2.1% | 100% | 4.7% | |
| 200/200 | N.A. | N.A. | 100% | 9.9% | 100% | 28.6% | |
| T-test exact MWSL | 50/50 | 84% | 1.4% | 86% | 0.8% | 98% | 1.3% |
| 100/100 | N.A. | N.A. | 100% | 3.8% | 100% | 8.2% | |
| 200/200 | N.A. | N.A. | 100% | 14.8% | 100% | 35.9% | |
| T-test general MWSL α′ = 1.5 × 10−6 | 50/50 | 72% | 0.8% | 74% | 0.4% | 96% | 0.6% |
| 100/100 | N.A. | N.A. | 100% | 2.7% | 100% | 6.1% | |
| 200/200 | N.A. | N.A. | 100% | 12.7% | 100% | 33.2% | |