Resumen de Selección genómica en el ganado vacuno de leche español
José Antonio Jiménez Montero, O. González Recio, Rafael Alenda Jiménez
This study includes initial stages for the implementation of genomic selection program in the Spanish Holstein population and has been used to obtain the official genomic breeding values and implement the imputation of genotypes. The objectives were: (1) To study alternative genotyping strategies for small populations, (2) to develop and validate methods for the evaluation of large data sets of genotypes, and (3) to study the effect of imputation process in predictive ability. The most informative strategy involved genotyping of females that exhibited upper and lower extreme values when compared with traditional pedigree index and a bull reference population. Including just top animals resulted in poor results. R-Boost resulted a competitive marker regression methodology in terms of predictive ability when compared with Bayes-A, Bayesian LASSO and G-BLUP. Optimization of the algorithm and behavior of tuning parameters were tested over real dairy cattle data reducing the time used with the original algorithm with negligible differences in accuracy and bias. Customized GoldenGate Bovine 3K and BovineLD BeadChip genotypes imputed to a BovineSNP50v2 BeadChip using Beagle achieved similar predictive ability than native genotypes. Subsequently, imputing to the BovineHD BeadChip and utilizing proper methods genomic predictions may be improved.
