Table 2. Predicted replication rates. We applied the replication analysis to phenotypes with at least two genome-wide significant variants in the discovery study. These phenotypes include body weight, albino coat color, agouti coat color, locomotor test day 1, and locomotor test day 2. We computed the true replication rate as the fraction of variants that were genome-wide significant in the discovery study that also passed the Bonferroni significance threshold in the replication study (“Empirical replication rate”). The model accounting for Winner’s Curse and confounding (“Predicted replication rate WC+C”) explains the true replication rate more accurately than the model accounting for only Winner’s Curse (“Predicted replication rate WC”).
| Discovery set | Replication set | Phenotype | Empirical replication rate | Predicted replication rate (WC) | Predicted replication rate (WC+C) |
|---|---|---|---|---|---|
| F34 GBS | F39-43 GBS | Body weight | 0.009 | 1.000 | 0.044 |
| Coat color, albino | 1.000 | 1.000 | 0.997 | ||
| Coat color, agouti | 0.932 | 1.000 | 0.577 | ||
| Locomotor test day 1 | 0.000 | 1.000 | 0.028 | ||
| Locomotor test day 2 | 0.000 | 1.000 | 0.140 | ||
| F39-43 GBS | F34 GBS | Body weight | 0.297 | 1.000 | 0.071 |
| Coat color, albino | 0.911 | 1.000 | 0.932 | ||
| Coat color, agouti | 0.815 | 1.000 | 0.925 | ||
| Locomotor test day 2 | 0.000 | 1.000 | 0.053 |