469
Comparison of Accuracies of Genomic Prediction in French Limousin Cattle Population according to the Number of Markers and to Pedigree Relationship between Training and Validation Populations

Monday, August 18, 2014
Posters (The Westin Bayshore)
Marine Barbat , UNCEIA, Jouy-en-Josas, France
Thierry Tribout , INRA UMR 1313 GABI, Jouy-en-Josas, France
Romain Saintilan , UNCEIA, Jouy-en-Josas, France
Eric Venot , INRA, UMR1313 GABI, Jouy-en-Josas, France
Marie-Noelle Fouilloux , Institut de l'Elevage - Idele, Jouy-en-Josas, France
Florence Phocas , INRA, UMR1313 GABI, Jouy-en-Josas, France
Abstract Text: Genomic breeding values were estimated in the French Limousin cattle breed for direct and maternal genetic effect traits routinely recorded at birth and weaning. A total of 1,646 bulls were genotyped with the Bovine SNP50 BeadChip® or the BovineHD BeadChip®. Their deregressed EBV were used in weighted analyses using a BayesC approach. Chip density and proportion of SNP with a non-zero effect did not affect the prediction accuracy. Accuracies of genomic EBV (GEBV) varied from 0.38 to 0.71 for direct effect traits and from 0.05 to 0.36 for maternal effect traits, when the validation animals had their sire in the training population. The GEBV of weakly related validation animals were on average 35% and 20% less accurate for direct and maternal effect traits, respectively. A prediction with only the 10,000 most important SNP resulted in similar GEBV accuracy than with all markers.

Keywords:

beef cattle

genomic selection