Table 3.
Power Comparisons of Sequence-Based and Variant-Based Replication under Fixed Effect Model
|
Ratesa |
Power for Replicationb |
|||||
|---|---|---|---|---|---|---|
| Number of Cases/Controls in Stage 1 and 2 Samples | False Positive | False Negative | Assay Success | Error Ratio | Sequence-Based | Variant-Based |
| 250/250c | 0 | 0 | 1 | 1 | 0.446 | 0.437 |
| 1% | 4% | 0.9 | 0.5 | 0.432 | 0.392 | |
| 1 | 0.386 | |||||
| 10% | 1% | 0.9 | 0.5 | 0.429 | 0.399 | |
| 1 | 0.390 | |||||
| 6.3% | 5% | 0.9 | 0.5 | 0.410 | 0.390 | |
| 1 | 0.378 | |||||
| 500/500c | 0 | 0 | 1 | 1 | 0.666 | 0.658 |
| 1% | 4% | 0.9 | 0.5 | 0.650 | 0.619 | |
| 1 | 0.607 | |||||
| 6.3% | 1% | 0.9 | 0.5 | 0.652 | 0.623 | |
| 1 | 0.613 | |||||
| 10% | 5% | 0.9 | 0.5 | 0.632 | 0.619 | |
| 1 | 0.600 | |||||
| 2000/2000d | 0 | 0 | 1 | 1 | 0.765 | 0.767 |
| 1% | 4% | 0.9 | 0.5 | 0.747 | 0.703 | |
| 1 | 0.689 | |||||
| 6.3% | 1% | 0.9 | 0.5 | 0.746 | 0.705 | |
| 1 | 0.694 | |||||
| 10% | 5% | 0.9 | 0.5 | 0.724 | 0.700 | |
| 1 | 0.669 | |||||
| 3500/3500d | 0 | 0 | 1 | 1 | 0.875 | 0.878 |
| 1% | 4% | 0.9 | 0.5 | 0.872 | 0.841 | |
| 1 | 0.834 | |||||
| 6.3% | 1% | 0.9 | 0.5 | 0.870 | 0.845 | |
| 1 | 0.835 | |||||
| 10% | 5% | 0.9 | 0.5 | 0.856 | 0.843 | |
| 1 | 0.825 | |||||
a
Significance levels for small-scale study: αS1 = 0.05 and αS2 = 0.05.
b
Significance levels for large-scale study: αS1 = 2.5 × 10−6 and αS2 = 2.5 × 10−6.
c
The impact of different combinations of false-positive/false-negative rate, assay success rate, and genotyping and sequencing error rate ratio on the replication power is examined.
d
The power was empirically estimated based upon 10,000 replicates.