TABLE 2.
Results of the SEM analysis on the simulated data
| Edge type | Model
|
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| IC | Measure | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | |
| BIC | F | FDR | 18.4 | 24.3 | 27.4 | 17.9 | 19.5 | 21.6 | 20.7 | 19.0 | 23.9 | 22.2 |
| Power | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 99.2 | 99.2 | 100.0 | 97.5 | 100.0 | ||
| B | FDR | 6.6 | 7.1 | 7.6 | 7.6 | 5.7 | 8.5 | 3.8 | 15.3 | 9.5 | 11.0 | |
| Power | 87.6 | 89.7 | 89.9 | 89.3 | 89.3 | 88.4 | 85.9 | 85.8 | 88.7 | 87.2 | ||
| BIC(δ) | F | FDR | 7.5 | 7.9 | 7.7 | 5.1 | 8.1 | 7.1 | 6.3 | 14.8 | 11.9 | 14.5 |
| Power | 100.0 | 100.0 | 99.2 | 99.2 | 100.0 | 96.7 | 100.0 | 98.4 | 100.0 | 100.0 | ||
| B | FDR | 0.8 | 0.0 | 1.7 | 0.0 | 1.6 | 3.4 | 0.0 | 3.4 | 1.8 | 0.9 | |
| Power | 80.7 | 82.2 | 79.9 | 78.5 | 81.2 | 76.2 | 77.9 | 77.7 | 72.7 | 71.8 | ||
False discovery rate and (percentage) power of edge detection are given for 10 artificial data sets using BIC and BIC(δ) criteria and separately for the eQTL → gene and gene → gene edges.