Open AccessMulti-varietal genomic selection in French pig populations
Author(s) -
Céline Carillier-Jacquin,
Llibertat Tusell,
Juliette Riquet,
MarieJosé Mercat,
Catherine Larzul
Publication year - 2016
Publication title -
hal (le centre pour la communication scientifique directe)
Language(s) - English
Resource type - Conference proceedings
DOI - 10.3920/978-90-8686-284-9
Subject(s) - selection (genetic algorithm) , genomic selection , computer science , biology , evolutionary biology , computational biology , genetics , artificial intelligence , genotype , gene , single nucleotide polymorphism
Reference population size is a limiting factor to succeed in genomic selection. Multi-breed genomic evaluations were proposed as a solution to increase reference population size but did not always lead to genomic breeding value (GEBV) accuracy improvement. The aim of this study is to investigate multi-varietal genomic evaluation using varieties of Piétrain pigs. The data consisted in a main reference population (V1) of 96 genotyped boars, their 752 genotyped and phenotyped offspring and two additional populations of 13 (V2) and 17 (V3) genotyped boars with their 118 and 177 genotyped and phenotyped offspring, respectively. We study the interest of adding the largest reference population to each of the two small populations on GEBV accuracies. Sixty traits were studied including production traits, skin lesions, hormonal and blood parameters. Single-step genomic evaluations were carried out either separately for each variety or by grouping the varieties (V1 and V2 or V1 and V3). Theoretical accuracies obtained in the single variety genomic evaluations were compared to the ones obtained in multi-varietal evaluations for the sires and the offspring. The breeding value predictability in single variety and multi-varietal genomic evaluations were also compared using a 4-fold cross-validation. Average genomic kinship coefficient between the main reference population and the other two populations was of 8% (V1 and V2) and 12% (V1 and V3). GEBV theoretical accuracies and predictive ability of animals from the smallest reference populations (V2 and V3) improved when considering multi-varietal evaluation. These improvements were higher for V2 which was most related to the main reference population. Slightly higher theoretical accuracies were obtained for the offspring than for the boars. This preliminary study suggests that increasing the reference population size with animals from another variety enhances the GEBV accuracy and the predictive ability of animals from small size populations. This enhancement is proportional to the degree of relatedness between varieties