473
Compression Efficiency Relationship Matrix: Developing New Methods to Determine Genomic Relationships for Improved Breeding

Monday, August 18, 2014
Posters (The Westin Bayshore)
Nicholas J Hudson , CSIRO, Brisbane, Australia
James Kijas , CSIRO Animal, Food and Health Sciences, Brisbane, Australia
Laercio R Porto-Neto , CSIRO Food Futures Flagship, Brisbane, Australia
Anthony Reverter-Gomez , Food Futures Flagship, CSIRO Animal, Food and Health Sciences, Brisbane, Australia
Abstract Text: Understanding genetic relatedness between individuals, sire groups and breeds underpins genomic selection and GWAS. Here, we describe a new estimate of genetic relatedness using normalized compression distance (NCD). Clustering of Sheep breeds inferred by NCD broadly reflects SNP correlation using standard multi-dimensional scaling. The clustering appears consistent with country of origin and population history. For example, the 4 British sheep meat breeds (Poll Dorset, Southdown, Suffolk and White Suffolk) clearly cluster with each other, but separate to unrelated breeds (Border Leicester, Merino and Texel). We show that the compression-based relationship matrix (CRM) and the genomic relationship matrix (GRM) are closely related. The quadratic relationship between pairwise NCD (CRM) and pairwise SNP correlation (GRM) implies CRM will perform better with closely related individuals, while the converse is true for GRM. For example, CRM resolves Merino from Poll Merino where GRM cannot.

Keywords: Genetic relationship matrix, Information compression, Sheep