Table 1. Realized and deterministic GBLUP accuracy with estimated heritabilities (h2) in bovine and constant Ne populations with either neutral QTL (Neut) or QTL under negative selection (Sel).
| Scenario | True h2 | Neut population estimated h2 (SE) | Sel population estimated h2 (SE) | Neut population realized accuracy (SE) | Sel population realized accuracy (SE) | Relative reduction in accuracy due to selection (%) | Deterministic prediction of accuracya |
|---|---|---|---|---|---|---|---|
| Constant SEQb | 0.45 | 0.459 (0.005) | 0.452 (0.001) | 0.595 (0.005) | 0.593 (0.005) | 0 | 0.662 |
| 0.15 | 0.161 (0.006) | 0.147 (0.006) | 0.427 (0.008) | 0.405 (0.007) | 5.1 | 0.420 | |
| Constant HD SNPb | 0.45 | 0.401 (0.006) | 0.371 (0.008) | 0.585 (0.006) | 0.562 (0.007) | 3.9 | 0.572 |
| 0.15 | 0.14 (0.005) | 0.120 (0.005) | 0.418 (0.009) | 0.378 (0.008) | 9.6 | 0.361 | |
| Bovine SEQb | 0.45 | 0.401 (0.005) | 0.413 (0.008) | 0.960 (0.001) | 0.961 (0.001) | 0 | 0.985 |
| 0.15 | 0.148 (0.005) | 0.145 (0.004) | 0.892 (0.003) | 0.888 (0.004) | 0.4 | 0.938 | |
| 0.01 | 0.010 (0.001) | 0.009 (0.001) | 0.591 (0.018) | 0.585 (0.013) | 1.0 | 0.624 | |
| Bovine HD SNPb | 0.45 | 0.440 (0.007) | 0.461 (0.010) | 0.961 (0.001) | 0.960 (0.001) | 0 | 0.983 |
| 0.15 | 0.154 (0.006) | 0.152 (0.005) | 0.894 (0.003) | 0.888 (0.004) | 0.7 | 0.935 | |
| 0.01 | 0.01 (0.001) | 0.009 (0.001) | 0.591 (0.018) | 0.586 (0.014) | 1.0 | 0.621 |
Standard errors across replicates are given in parentheses.
a
These accuracies are estimated using Equation 1 with our empirically estimated Me.
b
Accuracies are given for full sequence (SEQ) and high-density SNPs (HD SNPs) averaged across 20 replicates. The number of reference and validation individuals was 2500, except for the bovine scenario h2 = 0.01 where it was 3750.