Table 2 The prediction accuracies, biases (|1-b|) and mean squared errors (MSE) of different models under the default scenario across 20 replicates^a

Models ^b	Accuracies±s.e.	\|1-b\|±s.e.	MSE±s.e. (× 10 000)
Regular RR-TDM	0.50±0.01^A	0.08±0.01^A	37.77±1.32^A
GBLUP (DRP)	0.65±0.01^B	0.25±0.03^B	30.12±1.25^B
GBLUP (PGV)	0.65±0.01^B	0.24±0.01^B	31.47±1.25^C
SS RR-TDM	0.71±0.01^C	0.06±0.01^A	25.02±1.00^D

Abbreviations: DRP, de-regressed proof; GBLUP, genomic best linear unbiased prediction; PGV, predicted genetic value; SS RR-TDM, single-step random regression test-day model.
Common capital letters within columns donate no significant difference at the 0.01 level by paired t-test.
^aBias was measured as the deviation of regression coefficients (true additive genetic value on predicted genetic value) from 1. The results were from the default scenario with a heritability of 0.3, 100 QTL and random selection of sires in each generation.
^bModels applied were pedigree-based random regression test-day model (Regular RR-TDM), genomic best linear unbiased prediction (GBLUP) with response variables of predicted genetic value (GBLUP (PGV)) and de-regressed proof (GBLUP(DRP)), and single-step RR-TDM.

Quick links

Search