Strategies to implement genomic selection in Montana composite beef cattle

Author(s)

Sabrina Kluska

• Document type: Doctoral Thesis
• Press: Jaboticabal. 2021-01-20.
• Institution: Universidade Estadual Paulista (Unesp). Faculdade de Ciências Agrárias e Veterinárias. Jaboticabal
• Defense date: 2021-01-06
• Advisor: Fernando Sebastián Baldi Rey

Abstract

The search for new traits and methods able to increase the genetic gain in breeding programs is constant. With the advent of genomic selection, a great genetic progress has been observed, however the genomic selection in crossed animals, composite or multiracial populations is still not widespread. Thus, the aim of this study were: i) Investigate the impact of the use of different relationship matrices in the genetic evaluation of Montana cattle; ii) Investigate the effect of using metafounders (MFs) and unknown parents groups (UPG) on the genetic evaluation of Montana cattle. Records of 680,551 animals in the pedigree were available, of which 1,899 were genotyped with panels of different densities and subsequently imputed to the Neogen GeneSeek® Genomic Profiler (GGP) panel with around 30,000 SNPs. Phenotypic records of scrotal circumference at 12 months of age (SC12), postweaning weight gain (PWG), weaning weight (WW) and birth weight (BW) were available. Four distinct relationship matrices, and a single-trait model, were used: 1) Additive relationship matrix based on the pedigree (A); 2) Genomic relationship matrix, built as in the default ssGBLUP (G1); 3) Genomic relationship matrix, centered based on the specific-allele frequencies of the groups biological type groups or principal components (G2); 4) Genomic relationship matrix, centered and scaled based on the specific-allele frequencies of the groups of biological type or principal components (G3). In addition, metafounders and unknown parent groups were implemented in a multi-trait model at BLUP and ssGBLUP. Models with four or ten MFs and UPGs were tested. UPGs were added in the H matrix in ssGBLUP (ssGBLUP_UPG) or only in A^(-1) and A_22^(-1) (ssGBLUP_UPGA) matrices. For the validation, the method based on linear regression statistics (LR), and 436 animals with omitted phenotypes, were used. Adjustments in the genomic relationship matrices were not able to capture a greater proportion of additive genetic variance in relation to phenotypic variance, that is, to produce greater heritability. The addition of genomic information, in the model, in both studies, was able to increase the stability of the (G)EBVs. However, the stability of the (G)EBVs was higher when the single-trait model was used. All the comparison parameters used (stability of (G)EBVs, accuracy of models, BIF accuracy, dispersion, bias, mean of (G)EBVs and Spearman correlation) did not indicate any significant difference in predictions when the genomic relationship matrices were adjusted based on biological types or groups of principal components. Spearman's correlations of genetic values between pedigree-based and genomic models were low, indicating changes in the ranking of animals when selection is practiced with these models. However, when comparing the genomic models, with G adjusted or not, the correlation between the GEBVs was high, indicating little or no change in the classification of candidates for selection when the selection is made based on any of the genomic models. Overall, the inclusion of UPGs in the models produced a stability and dispersion similar to the other genomic models, however, biased genetic tendencies were observed when UPGs were considered only in the pedigree-based matrices. The inclusion of metafounders in the model was not able to produce considerable changes in the stability of (G)EBVs in two subsequent evaluations, except for PWG and dispersion of the models. However, the models with metafounders, either four or ten, produced a lower bias than the other genomic models, and similar to the BLUP based on the pedigree. These results indicate that the use of metafounders can reduce the bias of genomic predictions in Montana cattle to the same level as the pedigree-based models, with slightly less accuracy and correlation or stability of (G)EBV similar to the ssGBLUP default. (AU)

• FAPESP's process: 17/21573-0 - Genomic evaluation in composite cattle of the Montana breed using real and simulated data applying the BLUP genomic single-step method
• Grantee: Sabrina Kluska
• Support Opportunities: Scholarships in Brazil - Doctorate