Evaluating the benefits of including predictive SNP markers in single step evaluation in sheep using cross-validation

Abstract

<p>A SNP array of 50k SNP markers was used in single-step GBLUP (SS-GBLUP) models to estimate breeding values in the Australian sheep genetic evaluation system. In 2019, Neogen launched a new GeneSeek Genomic Profiler Ovine 50k chip, which included ~5000 SNPs that were identified based on Sheep CRC research as highly predictive for growth, carcass and eating quality traits. The objective of this work was to apply a five-fold cross-validation approach to compare different models for the use of predictive SNPs for post-weaning weight (PWT), carcass eye muscle depth (CEMD), carcass fat at C site (CCFAT), intramuscular fat (IMF) and shear force (SF5) based on the LAMBPLAN terminal sire genetic evaluation. Correlation and regression coefficients between adjusted phenotypes and SS-GBLUP EBVs for validation animals from the different models were calculated. The results indicated that adding predictive SNPs slightly improved the correlation and regression coefficient of EBVs, but there was no advantage in giving them more weight via a separate term in the model, confirming that the current industry evaluation model using a single genomic relationship matrix is the best of the tested models for these traits.</p>