45146

Genomic evaluation of eye pigmentation, hair coat and breed characterization in Hereford and Braford

Fernando Antonio Reimann ; Gabriel Soares Campos ; Leandro Lunardini Cardoso ; Arione A. Boligon ; Fernando F. Cardoso
Download
Favorite this paper

The present study aimed to estimate genetic parameters for a threshold animal model using traditional (pedigree and phenotype) and genomic (single-step, Bayes C?, Bayes B and Bayes A) methodologies, and measure accuracy gains for breeding value predictions with the incorporation of genomic information from complete SNPs panels or reduced panels containing most representative SNPs for eye pigmentation (PO), hair coat at weaning (PelD) and at yearling (PelS) and breed characterization (CR) traits in Hereford and Braford cattle. Phenotypic data from animals born between 1991 and 2014 belonging to the breeding program Delta G Connection were used, totaling 73,615 observations for PO, 81,043 for PelD, 41,390 for PelS and 28,186 for CR, along with pedigree data of 169 839 animals and genotypes of 3,954 animals and 41,011 SNP markers. Estimated heritabilities were of medium to high magnitude for PO (0,46?0,02; 0,46?0,02), PelD (0,44?0,03; 0,44?0,03)PelS (0.42 ? 0.02 ; 0.41 ? 0.02) and CR (0.33 ? 0.02, 0.32 ? 0.02) by traditional and genomic methods, respectively. Genetic correlations for all traits were of low magnitude (-0.24<rg<0.30), except between PelD and PelS (0.62). The use of genomic information resulted in increased accuracy levels by 28% and 14% for PO, 13% and 7% for PelD, 16% and 30% for PelS and 10% and 4% for CR, considering k-means and random grouping cross-validation, respectively. Moreover, using Genome-Wide Association (GWAS) approach, it was possible the detection of chromosome segments and SNP markers associated to these traits. Thus, very low-density SNP markers panels were proposed containing 159, 88, 122 and 40 markers for PO, PelD, PelS and CR, respectively. Such panels provided accuracy levels of 0.89 and 0.87, 0.77 and 0.87, 0.80 and 0.88, and, finally, 0.35 and 0.38, considering k-means and random strategies, respectively. It is possible to obtain genetic progress through selection for the studied traits, and genetic predictions were more accurate with the use of genomic information, particularly when using low-density panels of most representative SNPs.