Single-step BLUP (SS-BLUP) genomic prediction has the advantage of combining phenotypic information on both genotyped and nongenotyped animals. Recently developed single-step Bayesian regression models (SSBR) provide potentially even greater flexibility (e.g., heavy-tailed or variable selection alternatives) for modeling the prior distribution of SNP markers in tandem with polygenic effects. We discuss and present strategies for inferring upon these hyperparameters, particularly when data from either category (i.e., genotyped or nongenotyped) of animals is limiting. For example, when most animals in a genetic evaluation are not genotyped, inferences on key hyperparameters (i.e., marker variance, scale parameters) are compromised if these inferences are primarily based on marker data from genotyped animals only (strategy 1) whereas information on these hyperparameters might be readily borrowed from polygenic inferences involving nongenotyped animals as well (strategy 2). To compare these 2 strategies, a simulation study with 10 replicates was conducted. Five generations and a total of 2000 animals were simulated in each replicate. The heritability of the trait was 0.5 based on 1500 SNPs. The proportion of animals genotyped ranged from 10 to 90%. Results showed that strategy 2 estimated hyperparameters with greater accuracy than strategy 1 (Table 1), particularly when the proportion of animals that were genotyped was low (i.e., 10%). Nevertheless both strategies lead to similar accuracies of estimated breeding values for both genotyped and non-genotyped animals under the various genotyping rate scenarios. We also present methodology for both strategies using REML to infer upon these hyperparameters based on Gaussian specifications as well as how to infer upon hyperparameters in heavy-tailed (BayesA) and/or variable selection (BayesB) specifications. In conclusion, our proposed strategy had some advantage upon inferring hyperparameters, and further research is necessary for SSBR models. Table 1. Genetic variance by scenario and strategy

Scenario	Strategy	Total genetic variance(mean ± SD)
		24.60 ± 4.48 (True)
10% genotyped	1	24.92 ± 5.64
	2	25.70 ± 4.33
50% genotyped	1	22.35 ± 4.70
	2	24.05 ± 2.87
90% genotyped	1	20.73 ± 7.81
	2	24.26 ± 3.22