Avaliação de estratégias de genotipagem em situações de incerteza de paternidade e seu impacto sobre as avaliações genômicas em bovinos de corte
Carregando...
Arquivos
Data
2016-12-14
Autores
Tonussi, Rafael Lara [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Reprodutores múltiplos (RM) é o sistema de manejo mais comum em produção de bovinos de corte principalmente devido a facilidade e baixo custo de manejo. No entanto, RM não permite a identificação de paternidade, tornando o pedigree incompleto um dos principais obstáculos para avaliações genéticas precisas. Existe um crescente interesse na investigação do uso de dados genômicos em modelos com paternidade incerta, visando aumentar a acurácia e diminuir o viés nas avaliações genéticas. Portanto, o objetivo desse estudo foi investigar a aplicação do BLUP e single step genomic BLUP (ssGBLUP) em diferentes cenários com paternidade incerta utilizando dados simulados e reais em bovinos de corte. Foram simulados genótipos, pedigree e fenótipos para idade ao primeiro parto (IPP) e peso aos 550 dias (P550) utilizando herdabilidades baseadas em dados reais (0,12 para IPP e 0,34 para P550). O genoma foi simulado com um comprimento total de 2.333 cM, com 735.293 marcadores bialélicos e 7.000 QTL distribuídos aleatoriamente ao longo dos 29 cromossomos. Foi assumido que os QTLs explicaram 100% da variância genética. Para QTL, a quantidade de alelos variou aleatoriamente de dois a quatro por loci. Foram estudados cenários com 0%, 25%, 50%, 75% e 100% de RM e foram testados quatro tipos de escalas entre as matrizes G e A22. A acurácia e viés foram calculados para cincos grupos: ALL = todos os animais; BULL = apenas touros; GEN = animais genotipados, FEM = fêmeas e YOUNG = machos jovens. O uso da informação genômica no modelo (ssGBLUP) apresentou acurácia maior (variando de 0,31 a 0,97) do que o BLUP tradicional (variando de 0,05 a 0,97), especialmente no grupo YOUNG. No estudo com dados reais, todos os modelos incluíram grupos de contemporâneos e classe de idade da vaca como efeitos fixos. A acurácia do valor genético (EBV / GEBV) foi calculada em cada cenário para oito grupos de animais: ALL = todos os animais, BULL = apenas touros com dez ou mais progênies; GEN = animais genotipados, GENwithPHEN = animais genotipados com fenótipos, GENwithoutPHEN = animais genotipados sem fenótipos, YOUNG = machos e fêmeas jovens sem fenótipos, YwithoutGEN = animais jovens sem fenótipos e genótipos e YwithGEN = animais jovens sem fenótipos e com genótipos. As acurácias do EBV (método BLUP) variaram de 0,02 a 0,46 para P455 e 0,04 a 0,18 para IPP, enquanto que as acurácias do GEBV (ssGBLUP) variaram de 0,13 a 0,48 para P455 e 0,16 a 0,33 para IPP. Os resultados obtidos com dados simulados e reais mostraram que acurácias do EBV e GEBV diminuíram à medida que as proporções de RM aumentaram. Além disso, o uso da informação genômica na avaliação genética pelo ssGBLUP aumentou a acurácia da avaliação, especialmente para animais com menos informações e para animais jovens.
Multiple service sire (MS) is the most common mating system in extensive beef production systems mainly due the facility and low management cost. However, MS does not allow the paternity identification, which makes the incompleteness of pedigree one of the main obstacles for accurate genetic evaluations. There is a grown interest to investigate the use of genomic data in uncertain paternity models aiming to increase the accuracy and to decrease bias in genetic evaluations. Therefore, the objective of this study was to investigate the application of BLUP and single step genomic BLUP (ssGBLUP) under different scenarios of uncertain paternity, using simulated and real data in beef cattle population. Genotypes, pedigree, and phenotypes for age at first calving (AFC) and weight at 550 days (W550) were simulated using heritabilities based on real data (0.12 for AFC and 0.34 for W550). The simulated genome had a total length of 2,333 cM, with 735,293 biallelic markers and 7,000 QTLs randomly distributed over 29 BTA. It was assumed that QTLs explained 100% of the genetic variance. For QTL, the amount of alleles per loci ranged randomly from two to four. Uncertain paternity scenarios using 0%, 25%, 50%, 75%, and 100% were studied. Four ways of scaling the mean of the genomic matrix (G) to match the mean of the pedigree relationship matrix among genotyped animals (A22) were tested. Accuracy, bias and inflation were investigated for five groups of animals: ALL = all animals; BULL = only bulls; GEN = genotyped animals; FEM = females; YOUNG = young males. The use of genomic information in the model (ssGBLUP) provided more accurated prediction (ranging from 0.31 to 0.97) than traditional BLUP (ranging from 0.05 to 0.97), especially in the YOUNG group. In real data, all models included contemporary groups and age at calving in classes as fixed effects. The accuracy of the estimated breeding value (EBV/GEBV) prediction was calculated in each scenario with eight groups of animals: ALL = all animals in the population, BULL = only bulls with ten or more progenies; GEN = genotyped animals, GENwithPHEN = genotyped animals with phenotypes, GENwithoutPHEN = genotyped animals without phenotypes, YOUNG = male and female young animals without phenotypes, YwithoutGEN = young animals without phenotypes and genotypes, and YwithGEN = young animals without phenotypes and with genotypes. Accuracies of EBV (BLUP method) ranged from 0.02 to 0.46 for W450 and 0.04 to 0.18 for AFC, while the accuracies of GEBV (ssGBLUP) ranged from 0.13 to 0.48 for W450 and 0.16 to 0.33 for AFC. The results obtained in simulated and real data showed that EBV and GEBV accuracy decreased as the proportion of MS increased. Additionally, the use of genomic information in the genetic evaluation by ssGBLUP increases the accuracy of evaluation, especially for animals with few number of information, such as young animals.
Multiple service sire (MS) is the most common mating system in extensive beef production systems mainly due the facility and low management cost. However, MS does not allow the paternity identification, which makes the incompleteness of pedigree one of the main obstacles for accurate genetic evaluations. There is a grown interest to investigate the use of genomic data in uncertain paternity models aiming to increase the accuracy and to decrease bias in genetic evaluations. Therefore, the objective of this study was to investigate the application of BLUP and single step genomic BLUP (ssGBLUP) under different scenarios of uncertain paternity, using simulated and real data in beef cattle population. Genotypes, pedigree, and phenotypes for age at first calving (AFC) and weight at 550 days (W550) were simulated using heritabilities based on real data (0.12 for AFC and 0.34 for W550). The simulated genome had a total length of 2,333 cM, with 735,293 biallelic markers and 7,000 QTLs randomly distributed over 29 BTA. It was assumed that QTLs explained 100% of the genetic variance. For QTL, the amount of alleles per loci ranged randomly from two to four. Uncertain paternity scenarios using 0%, 25%, 50%, 75%, and 100% were studied. Four ways of scaling the mean of the genomic matrix (G) to match the mean of the pedigree relationship matrix among genotyped animals (A22) were tested. Accuracy, bias and inflation were investigated for five groups of animals: ALL = all animals; BULL = only bulls; GEN = genotyped animals; FEM = females; YOUNG = young males. The use of genomic information in the model (ssGBLUP) provided more accurated prediction (ranging from 0.31 to 0.97) than traditional BLUP (ranging from 0.05 to 0.97), especially in the YOUNG group. In real data, all models included contemporary groups and age at calving in classes as fixed effects. The accuracy of the estimated breeding value (EBV/GEBV) prediction was calculated in each scenario with eight groups of animals: ALL = all animals in the population, BULL = only bulls with ten or more progenies; GEN = genotyped animals, GENwithPHEN = genotyped animals with phenotypes, GENwithoutPHEN = genotyped animals without phenotypes, YOUNG = male and female young animals without phenotypes, YwithoutGEN = young animals without phenotypes and genotypes, and YwithGEN = young animals without phenotypes and with genotypes. Accuracies of EBV (BLUP method) ranged from 0.02 to 0.46 for W450 and 0.04 to 0.18 for AFC, while the accuracies of GEBV (ssGBLUP) ranged from 0.13 to 0.48 for W450 and 0.16 to 0.33 for AFC. The results obtained in simulated and real data showed that EBV and GEBV accuracy decreased as the proportion of MS increased. Additionally, the use of genomic information in the genetic evaluation by ssGBLUP increases the accuracy of evaluation, especially for animals with few number of information, such as young animals.
Descrição
Palavras-chave
Acurácia, Avaliação genética, Pedigree incompleto, Reprodutores múltiplos, Viés, Accuracy, Genetic evaluation, Missing pedigree, Multiple sires, Bias