SNP-Panel Design for Dairy Proportion Estimation and Parentage Testing

Abstract

The selection of small numbers of SNPs to analyse population features is an important task in the livestock industry. Populations differ in their genetic architecture, which often requires the selection of population specific SNPs. Different tasks, such as breed proportion prediction or parentage testing, also require specific panels. We tested which selection methods are best for breed proportion estimation and parentage testing in a crossbred dairy population from East Africa. We selected SNPs from a 735k SNP panel (Illumina) based on several methods:a) high minor allele frequencies; b) high allele frequency differences between ancestral populations; c) at random; d) with a differential evolution algorithm. Estimates of breed proportions in the subsets were tested against true breed proportions based on all 770k SNP obtained from ADMIXTURE. Parentage assignments was based on opposing homozygotes. Panels selected for largest allele frequency differences in ancestral populations gave best results for breed proportion predictions and panels selected for highest minor allele frequency gave best parentage resolution.