A Binary Classifier Using SNP Data for Prediction of Phenotypic Outcomes in Hanwoo (Korean) Cattle

Abstract

Korean 'Hanwoo' cattle are prized for their high marbling ability and meat quality. Classically, these cattle possess a homogeneous yellow coat colouring, with farmers believing that 'Hanwoo' with white spotted coats are crossbred and therefore unacceptable for breeding purposes. In this study we first attempted to determine if the coat spots were due to a non-'Hanwoo' genetic background or, alternatively, if the trait is intrinsic to the breed. By genotyping 232 (136 spotted) animals from half-sib families on the Illumina Bovine 50K SNP array, we compared the genotyped Hanwoo to other unrelated 'Hanwoo' and European taurine breeds using principal component analysis. Results showed no evidence of crossbreeding in the spotted animals. A differential evolution algorithm was then used to evolve a classifier for the trait which selected 12 SNP with an accuracy of ~82% in separating individuals; further investigation using only haplotypes inherited from the sires resulted in a marked improvement to ~92% accuracy for these 12 SNP. This research highlights the potential for using these SNP as genetic markers to either entirely remove the trait from the population in the long term or manage matings so that the trait is not expressed in the offspring.