Assessment strategies genotyping in situations of uncertainty of paternity and its impact on the evaluation genomic in beef cattle

Abstract

The multiple sires (MS) system is the mating system most used for beef cattle, considering the few studies of Nellore breed and the importance of genomic selection, it is necessary studies which evaluate the inclusion of models with animal that has uncertain of paternity and the creation of the pedigree with genomic data in different scenarios and verify the impact on genetic evaluations. Thus, the objective of this work is evaluate the impact application of the genomic information in situations which uncertain of paternity and different strategies for genotyping on beef cattle genetic evaluation using real and simulated data. For the simulation study, a relationship matrix (A) will be created and from it scenarios with 50, 75 and 100% of parents (bulls) known will be created. In relation the creation of the G matrix, it will be tested strategies for genotyping animals with different proportions of progeny of MS in each scenarios. Therefore, several scenarios with various models with different proportions of known parents in the matrix A will be tested, for two traits, with low and moderate heritability, considering two cases (50 and 100% known parents) of genotyped animal with uncertain of paternity. The models will be used in each situation are: BLUP, animal hierarchical model, multiple and single-step. For the study with real data will be evaluated two situations where the first genomic information will be included for all animals and the second will be included only genotypes of animals with certain paternity. In the simulation study the correlation between TBV (true breeding value) and EBV (estimated breeding value) will be calculated to evaluate the prediction hability of the models. To estimate the bias estimates, the true genetic value is regressed on the respective estimated breeding value for each scenario. In the study with real genotypic and phenotypic data, the prediction hability of the proposed models will be tested through cross-validation. (AU)