Table 4. Accuracy and optimal parameters of BLUP|GA for common datasets obtained from the training stage.

Dataset	Trait	Accuracy		BLUP|GA Parameters
		GBLUP	BLUP|GA	top%a	weightb	nflankc
Loblolly pine	Rustbin	0.298	0.385d	0.12	0.140	0
	Gall	0.237	0.346	0.32	0.450	0
	Density	0.238	0.241	5.00	0.024	0
	Rootnum	0.268	0.270	5.20	0.024	0
	CWAC	0.475	0.478	0.15	0.006	0
	Rootnum_bin	0.288	0.288	0.25	0.005	0
QTL-MAS2012	T1	0.707	0.779	0.40	0.280	5
	T2	0.717	0.802	0.20	0.300	5
	T3	0.761	0.847	0.20	0.600	5
GSA dataset	PolyUnres	0.453	0.454	5.00	0.010	2
	GammaUnres	0.442	0.546	0.12	0.123	2
	PolyRes	0.390	0.391	6.00	0.010	2
	GammaRes	0.410	0.491	0.17	0.175	2

BLUP|GA, best linear unbiased prediction-given genetic architecture; GBLUP, genomic best linear unbiased prediction; GSA, Genetics Society of America; SNP, single-nucleotide polymorphism.

^aPercentage of top SNPs.

^bOverall weight ω for the genetic architecture part while building T matrix.

^cNumber of selected flanking SNPs near each top SNPs, the nflank was set to 0 for Loblolly and not chosen in a validation procedure.

^dScenario with the highest accuracy is shown in bold face.