454
Improving predictive ability of selected subsets of single nucleotide polymorphisms in a moderately sized dairy cattle population
Three methods were tested to select subsets of SNPs from the BovineSNP50 BeadChip to compute genomic evaluations for dairy cattle populations with ~1000 genotyped bulls: (1) their effects on genetic evaluations of bulls with daughter records for protein production in 2009 through 2013; (2) their effects in the 2009 evaluations, also including bulls without daughter records; and (3) the regression of the SNPs allelic frequency on the bulls' birth dates. The optimum number of SNPs ranged from 800 to 4000. Correlations for method 1 of 0.8 between genomic and 2013 evaluations for validation bulls were obtained if markers were selected based on 2013 evaluations. Correlations of genomic with current evaluations for method 2 were greater than parent average correlations for most traits analyzed, and means were generally less biased than parent averages. Thus method 2 is optimal for populations of this size.
Keywords:
genomic evaluations
dairy cattle
genetics
