Table 2. Precision and bias of potency metrics in gain-loss models.
| Data Set Parameters | Estimator | |||||
|---|---|---|---|---|---|---|
| True |RMAX| | True AC50(G) | True AC50(L) | True PODWES | True Peak | AC50 | PODWES |
| 5% error (15% Detection Limit) | ||||||
| 25 | 0.001 | 0.1 | 3.23e-03 | 20.5 | 2.41 (+2.71) | 1.44 (+0.12) |
| 25 | 0.001 | 10 | 3.17e-03 | 24.5 | 7.48 (+2.03) | 1.32 (+0.11) |
| 25 | 0.1 | 10 | 0.36 | 20.5 | 3.28 (−0.82) | 1.37 (−0.07) |
| 50 | 0.001 | 0.1 | 1.43e-03 | 40.95 | 0.85 (+2.51) | 1.03 (−0.04) |
| 50 | 0.001 | 10 | 1.35e-03 | 49.0 | 6.42 (+1.95) | 1.03 (−0.03) |
| 50 | 0.1 | 10 | 0.18 | 40.95 | 0.88 (−0.51) | 1.14 (−0.27) |
| 100 | 0.001 | 0.001 | 5.80e-04 | 24.3 | 1.32 (+1.15) | 0.86 (+0.05) |
| 100 | 0.001 | 0.1 | 4.27e-04 | 81.9 | 0.37 (+2.48) | 1.07 (+0.05) |
| 100 | 0.001 | 10 | 4.19e-04 | 98.0 | 6.44 (+1.98) | 1.07 (+0.06) |
| 100 | 0.1 | 0.1 | 0.08 | 25.0 | 19.78 (−0.29) | 0.95 (−0.20) |
| 100 | 0.1 | 10 | 0.07 | 81.9 | 0.37 (−0.48) | 1.21 (−0.16) |
| 100 | 10 | 10 | 9.47 | 24.3 | 1.36 (−1.17) | 0.94 (−0.42) |
| 10% error (30% Detection Limit) | ||||||
| 50 | 0.001 | 0.1 | 2.55e-03 | 40.95 | 2.60 (+2.72) | 1.44 (+0.17) |
| 50 | 0.001 | 10 | 2.89e-03 | 49.0 | 7.33 (+1.96) | 1.32 (+0.12) |
| 50 | 0.1 | 10 | 0.31 | 40.95 | 3.47 (−0.80) | 1.36 (−0.04) |
| 100 | 0.001 | 0.1 | 1.37e-03 | 81.9 | 0.87 (+2.51) | 1.00 (−0.06) |
| 100 | 0.001 | 10 | 1.25e-03 | 98.0 | 6.42 (+1.95) | 1.00 (−0.03) |
| 100 | 0.1 | 10 | 0.16 | 81.9 | 0.87 (−0.51) | 0.94 (−0.32) |
Equation (2) in the Methods was used to simulate data sets with the specified error. Log10Precision is shown with log10Bias in parentheses. Only data sets with true responses above the assay detection limit were evaluated.