Comparison of Genome Wide Association Studies for milk production traits in Valle del Belice dairy sheep

The great achievements in livestock species selection during the last 50 years largely relied on quantitative genetic theory and infinitesimal genetic model. In the last 20 years, due to the application of advanced techniques in molecular genetics and statistics, several chromosomal regions that influence quantitative traits have been discovered. Combinations of molecular and classical quantitative information in a composite selection index have been proposed to increase the accuracy of selection. Nowadays, many genotyping arrays for thousands of SNPs are available for several livestock species, such as: cattle, sheep, pigs, horses, goat and chickens. The overall aim of this thesis is the comparison of different GWAS approaches to identify SNPs associated with milk production traits in Valle del Belice dairy sheep. In particular different genetic merit indices (breeding values and their deregressed and weighted values) and single test day observations will be evaluated to identify SNPs associated with milk yield (MY), fat yield (FY), fat percentage (F%), protein yield (PY) and protein percentage (P%) using different statistical approaches. The raw phenotypes data set included 5,586 observations of 481 ewes for MY, F%, FY, P% and PY traits. All animals were genotyped using the Illumina OvineSNP50K BeadChip. A single trait repeatability test-day animal model was performed to estimate the breeding values. The EBVs for MY, FY, F%, PY and P% estimated with the mixed model were deregressed (DEBV) and weighted (DEBVw) to obtain a more accurate estimate of the expected phenotype according to Garrick et al. (2009). Genome-wide association analysis was carried out based on regression of phenotypes (EBVs, DEBVs and DEBVw) with the genotypes of animals for one SNP at a time. For single-marker GWAS, we used a three-step approach referred to as genomic GRAMMAR-GC as implemented in GenABEL package. Other two approaches for genome wide association study were used. The first accounts the covariances between repeated measures for each individual using the RepeatABEL package was used. The last approach known as Regional Genomic Relationship Mapping or Regional Heritability Mapping (RHM) provides heritability estimates attributable to small genomic regions, and it has the power to detect regions containing multiple alleles that individually contribute too little variance to be detected by standard GWA studies. Comparison among the estimated breeding values and their deregressed and weighted values as responsible variables, respect to their influence on our GWAS results, has demonstrated that DEBVs and DEBVw allow identifying a greater numbers of SNPs than using EBVs. Several SNPs using different approaches were identified and some of these SNPs are mapped within the previously reported QTL regions and within candidate genes for milk production traits. The results confirmed the roles of LALBA gene and AQP genes, on OAR 3, as candidate genes for milk production traits in sheep. Moreover some genomic regions identified by close SNPs associated with a specific trait should be further investigated to verify their effect on the traits. The general consistence of the significant SNPs detected herein with the reported QTL and candidate genes for milk traits allow us to be confident of the results obtained. The information generated from this thesis has important implications for the design and applications of association studies as well as for the development of selection breeding programs for the Valle del Belice sheep breeds.