Abstract

Genome-wide estimated breeding values can be computed from the simultaneous estimates of the effects of small intervals of DNA throughout the genome on a trait or traits of interest. Small intervals or segments of DNA can be created by the use of thousands of single nucleotide polymorphisms (SNP) available in panels of 10, 25 and 50 thousand SNP. A simulation study was conducted to compare factors that could influence the accuracy of genome-wide selection. Factors studied were the heritability of the trait, dispersion of quantitative trait loci (QTL) across the genome and size of the QTL effects. A 100-cM genome was assumed with 100 equally spaced SNP markers and 10 QTL. A granddaughter design was constructed with 20 sires and 100 sons per sire. Population-wide linkage disequilibrium was assumed to be sufficient after 25 generations of random mating starting with 30 sires and 400 dams. Best linear unbiased prediction was used to simultaneously estimate the effects of 99 SNP intervals, based on determining the SNP haplotype of each son inherited from the sire. Indicator variables were used in the model to indicate haplotype transmission. A genome-wide estimated breeding value was calculated as the sum of the appropriate haplotype interval estimates for each son. Correlations between estimated and true breeding values ranged from 0.60 to 0.79. Situations with unequally sized QTL effects and randomly dispersed QTL gave higher correlations. QTL positions could be estimated to within 2 cM or less.