Table 2.

Summary of rrBLUP-based GWAS-assisted genomic prediction models of STB resistance scored in 175 winter wheat genotypes.

Number of markers set as fixed effects	Type of marker selection for fixed effects
	Markers selected by significance in GWAS		Completely random selection of markers
	Average model accuracy	95% confidence interval of the mean	Average model accuracy	95% confidence interval of the mean
0	0.47	N/A	N/A	N/A
1	0.48	[0.44, 0.51]	0.44	[0.43, 0.44]
2	0.51	[0.49, 0.53]	0.44	[0.43, 0.45]
3	0.54	[0.52, 0.56]	0.45	[0.42, 0.48]
4	0.58	[0.55, 0.61]	0.43	[0.41, 0.45]
5	0.62	N/A	0.44	[0.41, 0.47]

The models utilized permutations of 1 to 5 markers in significant association with STB resistance identified in the same population. The models were compared against a model containing no fixed effects and a series of models that sampled equally sized subsets of random markers, where each subset of random markers was repeated five times. All models were validated against the same set of 80/20 training/test sets (N = 500). The zero and five GWAS-selected marker models were only repeated once, and thus have no confidence interval data.