The inclusion of genotype information in several cattle breeding programmes all around the world lead to the genomic selection (GS) era. One of the main advantages of GS over the traditional selection is the possibility of an early estimation of breeding values of candidates based on their genotypes. The optimisation of genotyping and phenotyping strategy represents a key point for GS programmes. Aim of this study was to evaluate the effect of using different phenotyping and genotyping strategies on sire breeding values (BV) accuracies. Using QMSim a dairy bovine population was simulated. Five replicates of ten recent populations were simulated starting from 200 males and 50,000 females. Best animals were selected and mated using a positive assortative mating design; sire and dam replacement were fixed at 40 and 30%, respectively. Ten identical chromosomes each with 1000 markers were generated for the last three generations. About 26,000 females per generation were available. The phenotypes of the last generation were masked in order to represent the candidates of a breeding programmes. Using blupf90 family programmes, breeding values were estimated for 160 sires of female belonging to the tenth generation. Some of those sires had daughters also in the ninth generation. BV accuracies were computed as root square of reliabilities. Masking phenotypic and genetic information of candidate females, the average BV accuracy of sires was 0.54±0.48. The high standard deviation can be ascribed to the different number of daughters in the previous generations. Increases were observed when phenotypes or genotypes were added in the analyses: 0.92±0.02 and 0.75±0.15 for phenotypes and genotypes, respectively. The best scenario was found using both phenotypes and genotypes, with an average accuracy of 0.93±0.01. Using phenotypes in lieu of genotypes lead to higher BV accuracies. However, the phenotypes registration implies longer generation intervals and increase of costs. The quite good accuracies highlighted using only genotypes and the low genotyping price at this point can suggest the use of this strategy instead of waiting for phenotypes collection also for the well-known advantage of having BV already available at candidate birth. As expected, the highest accuracies were found when all possible sources of information (phenotypes+genotypes) were included in the model. [Ital J Anim Sci vol.18:s1, 2019] [page 91] Italian Journal of Animal Science 2019; volume
Genomic breeding values accuracies using phenotypes or genotypes.
Giustino Gaspa
2019-01-01
Abstract
The inclusion of genotype information in several cattle breeding programmes all around the world lead to the genomic selection (GS) era. One of the main advantages of GS over the traditional selection is the possibility of an early estimation of breeding values of candidates based on their genotypes. The optimisation of genotyping and phenotyping strategy represents a key point for GS programmes. Aim of this study was to evaluate the effect of using different phenotyping and genotyping strategies on sire breeding values (BV) accuracies. Using QMSim a dairy bovine population was simulated. Five replicates of ten recent populations were simulated starting from 200 males and 50,000 females. Best animals were selected and mated using a positive assortative mating design; sire and dam replacement were fixed at 40 and 30%, respectively. Ten identical chromosomes each with 1000 markers were generated for the last three generations. About 26,000 females per generation were available. The phenotypes of the last generation were masked in order to represent the candidates of a breeding programmes. Using blupf90 family programmes, breeding values were estimated for 160 sires of female belonging to the tenth generation. Some of those sires had daughters also in the ninth generation. BV accuracies were computed as root square of reliabilities. Masking phenotypic and genetic information of candidate females, the average BV accuracy of sires was 0.54±0.48. The high standard deviation can be ascribed to the different number of daughters in the previous generations. Increases were observed when phenotypes or genotypes were added in the analyses: 0.92±0.02 and 0.75±0.15 for phenotypes and genotypes, respectively. The best scenario was found using both phenotypes and genotypes, with an average accuracy of 0.93±0.01. Using phenotypes in lieu of genotypes lead to higher BV accuracies. However, the phenotypes registration implies longer generation intervals and increase of costs. The quite good accuracies highlighted using only genotypes and the low genotyping price at this point can suggest the use of this strategy instead of waiting for phenotypes collection also for the well-known advantage of having BV already available at candidate birth. As expected, the highest accuracies were found when all possible sources of information (phenotypes+genotypes) were included in the model. [Ital J Anim Sci vol.18:s1, 2019] [page 91] Italian Journal of Animal Science 2019; volume
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
