Table 4.

Accuracy of genomic prediction (standard deviation) under different subsets of F_ST based and randomly selected SNPs for populations^a P₁ and P₂ (average over 5 replicates)

	Accuracyb
	P1 (LD = 0.3)	P2 (LD = 0.7)
FST based
2.5 K	0.724 (0.021)	0.805 (0.014)
5 K	0.736 (0.022)	0.823 (0.012)
10 K	0.740 (0.023)	0.828 (0.013)
20 K	0.741 (0.027)	0.824 (0.013)
40 K	0.735 (0.027)	0.815 (0.014)
80 K	0.728 (0.028)	0.802 (0.012)
160 K	0.723 (0.031)	0.784 (0.013)
Random
2.5 K	0.600 (0.054)	0.669 (0.019)
5 K	0.640 (0.047)	0.709 (0.015)
10 K	0.676 (0.036)	0.736 (0.019)
20 K	0.695 (0.037)	0.746 (0.014)
40 K	0.707 (0.034)	0.754 (0.010)
80 K	0.712 (0.033)	0.757 (0.013)
160 K	0.715 (0.031)	0.759 (0.011)
Full panel
400 K	0.716 (0.032)	0.760 (0.011)

^aP₁: 200 QTLs and linkage disequilibrium (LD) between adjacent markers equal to 0.3 and P₂: 200 QTLs and LD between adjacent markers equal to 0.7;^b correlation between true and predicted breeding values