Table 5.
Average (standard error) of Mean Square Error of Prediction (MSEP) of MEBV for individuals in the evaluation population.
| Method | unequal QTL variance | equal QTL variance | ||||
|---|---|---|---|---|---|---|
| low nQTL | int. nQTL | high nQTL | low nQTL | int. nQTL | high nQTL | |
| sc. 1 | sc. 2 | sc. 3 | sc. 4 | sc. 5 | sc. 6 | |
| BM | 659 (26) | 4049 (108) | 10463 (343) | 79 (2) | 416 (6) | 850 (12) |
| LARS | 707 (24) | 4019 (71) | 8230 (124) | 91 (2) | 465 (6) | 927 (12) |
| PLSR | 993 (24) | 4242 (73) | 8405 (123) | 93 (2) | 458 (6) | 922 (14) |
Methods BM, LARS and PLSR were used to calculate the MEBV. The simulated number of QTL was low (low nQTL), intermediate (int. nQTL) or high (high nQTL). The simulated variance of every tenth QTL was 81 times larger than variance of remaining QTL (unequal QTL variance) or equal for all QTL (equal QTL variance). The averages and standard deviations were calculated using 60 replicated simulations.