Table 2.
Posterior mean of (1-π) multiplied by K = 2,000 loci used in the analyses (se) according to the Bayesian method, number of QTL, distribution of QTL effects and training data size.
| QTL effect distribution and training data size | |||||
|---|---|---|---|---|---|
| Gamma | Normal | ||||
| Method | No. of QTL | 1,000 | 4,000 | 1,000 | 4,000 |
| BayesCπ | 10 | 7 (1) | 7 (0.8) | 13 (0.9) | 12 (0.8) |
| 200 | 69 (5) | 86 (3) | 236 (13) | 204 (3) | |
| 1,000 | 312 (40) | 315 (8) | 1,230 (91) | 1,007 (19) | |
| BayesDπ | 10 | 165 (11) | 59 (3) | 229 (9) | 81 (4) |
| 200 | 645 (22) | 343 (7) | 952 (24) | 564 (6) | |
| 1,000 | 984 (39) | 747 (10) | 1,169 (36) | 1,227 (12) | |
The starting value for π was 0.5. Results are based on 24 replicates of the ideal simulation in which all loci were in linkage equilibrium and both SNPs and QTL were modeled