Strategies for within-litter selection of piglets using ultra-low density SNP panels

Abstract

Genotyping costs and the large number of selection candidates are major factors that inhibit the application of genomic selection in the swine industry and other small-sized livestock species. In order to reduce genotyping costs and increase the uptake of genomic selection, a possible strategy is to genotype animals with an affordable low-density (LD) SNP panel and, then accurately impute the LD panel to a high-density (HD) SNP panel. For within-litter piglet selection, genotyping all piglets from all farrows using the commercially available SNP chips is still cost prohibitive. Consequently, genomic evaluation is limited in this stage and genotypic and phenotypic data from all piglets in a litter are rarely available. This study investigates the feasibility of implementing genomic selection for within-litter piglet selection, using a total of nine simulated LD panels: from the “ultra” low (300–3000 SNP markers) to moderately low (6000–10, 000 SNP markers). For each LD panel, the performance of the genomic predictions according to the accuracy of genotype imputation, the accuracy of the genomic estimated breeding values (GEBV) based on the imputed data, and distribution of the correctly selected animals within litter was evaluated and compared to using the simulated HD panel (60,000 SNP) and True Breeding Values (TBVs). In this simulation study, we considered three economically important traits: back fat thickness (BF), growth rate of age to 100 Kg (GR), and litter size (LS). For the LD panel sizes ranging from 300 to 10,000, the accuracy of imputation (measured as concordance rate) ranged from 73.20 to 99.81%; and the mean proportion of the correctly selected top rank animals within litter ranged from 55 to 98%. Based on the trade-off between panel size and genomic selection accuracy, the use of a LD panel containing 1500 SNPs might be recommended, as this panel yielded more than 85% correctly selected animals within-litter based on all three traits.