TABLE 3.
Accuracy of GEBVs with high-density (HD) genotyping and loss of accuracy using approaches based on genotyping evenly spaced low-density (ELD) SNPs, as a percentage of the accuracy of HD-BayesB (HD-BB) in generations after training and for a trait with 100, 500, and 1000 simulated QTL
| No. of simulated QTL | Generation
|
|||
|---|---|---|---|---|
| Method | 2 | 3 | 4 | |
| HD-BB | 70.9 (±1.64) | 69.0 (±1.55) | 65.1 (±1.96) | |
| 100 | ELD-10 | 4.4 (±3.16) | 9.5 (±3.50) | 15.8 (±4.60) |
| ELD-10+ | 4.4 (±3.16) | 5.0 (±3.08) | 4.5 (±3.63) | |
| HD-BB | 64.4 (±1.03) | 58.2 (±1.41) | 55.3 (±1.37) | |
| 500 | ELD-10 | 4.4 (±3.29) | 8.6 (±4.82) | 13.9 (±5.43) |
| ELD-10+ | 4.4 (±3.29) | 5.7 (±3.89) | 5.2 (±3.71) | |
| HD-BB | 63.2 (±1.20) | 57.1 (±1.69) | 56.1 (±1.79) | |
| 1000 | ELD-10 | 4.6 (±2.98) | 8.6 (±3.81) | 14.2 (±4.90) |
| ELD-10+ | 4.6 (±2.98) | 4.5 (±3.12) | 6.0 (±3.86) | |
One thousand individuals each having 1000 HD-SNPs were used to estimate HD-SNP effects with BayesB and LD-SNPs at intervals of 10 cM were used to trace HD-SNP alleles (±SE, based on 25 replicates). ELD-10+ indicates that individuals used for breeding in each generation were also genotyped for HD-SNPs.