Genomic prediction ability for feed efficiency traits using different models and pseudo-phenotypes under several validation strategies in Nelore cattle.

There is a growing interest to improve feed efficiency (FE) traits in cattle. The genomic selection was proposed to improve these traits since they are difficult and expensive to measure. Up to date, there are scarce studies about the implementation of genomic selection for FE traits in indicine cattle under different scenarios of pseudo-phenotypes, models, and validation strategies on a commercial large scale. Thus, the aim was to evaluate the feasibility of genomic selection implementation for FE traits in Nelore cattle applying different models and pseudo-phenotypes under validation strategies. Phenotypic and genotypic information from 4 329 and 3 467 animals were used, respectively, which were tested for residual feed intake, DM intake, feed efficiency, feed conversion ratio, residual BW gain, and residual intake and BW gain. Six prediction methods were used: single-step genomic best linear unbiased prediction, Bayes A, Bayes B, Bayes Cπ, Bayesian least absolute shrinkage and selection operator (BLASSO), and Bayes R. Phenotypes adjusted for fixed effects (Y*), estimated breeding value (EBV), and EBV deregressed (DEBV) were used as pseudo-phenotypes. The validation approaches used were: (1) random: the data was randomly divided into ten subsets and the validation was done in each subset at a time; (2) age: the partition into training and testing sets was based on year of birth and testing animals were born after 2016; and (3) EBV accuracy: the data was split into two groups, being animals with accuracy above 0.45 the training set; and below 0.45 the validation set. In the analyses that used the Y* as pseudo-phenotype, prediction ability (PA) was obtained by dividing the correlation between pseudo-phenotype and genomic EBV (GEBV) by the square root of the heritability of the trait. When EBV and DEBV were used as the pseudo-phenotype, the simple correlation of this quantity with the GEBV was considered as PA. The prediction methods show similar results for PA and bias. The random cross-validation presented higher PA (0.17) than EBV accuracy (0.14) and age (0.13). The PA was higher for Y* than for EBV and DEBV (30.0 and 34.3%, respectively). Random validation presented the highest PA, being indicated for use in populations composed mainly of young animals and traits with few generations of data recording. For high heritability traits, the validation can be done by age, enabling the prediction of the next-generation genetic merit. These results would support breeders to identify genomic approaches that are more viable for genomic prediction for FE-related traits.

Genome-enabled prediction of reproductive traits in Nellore cattle using parametric models and machine learning methods.

This study aimed to assess the predictive ability of different machine learning (ML) methods for genomic prediction of reproductive traits in Nellore cattle. The studied traits were age at first calving (AFC), scrotal circumference (SC), early pregnancy (EP) and stayability (STAY). The numbers of genotyped animals and SNP markers available were 2342 and 321 419 (AFC), 4671 and 309 486 (SC), 2681 and 319 619 (STAY) and 3356 and 319 108 (EP). Predictive ability of support vector regression (SVR), Bayesian regularized artificial neural network (BRANN) and random forest (RF) were compared with results obtained using parametric models (genomic best linear unbiased predictor, GBLUP, and Bayesian least absolute shrinkage and selection operator, BLASSO). A 5-fold cross-validation strategy was performed and the average prediction accuracy (ACC) and mean squared errors (MSE) were computed. The ACC was defined as the linear correlation between predicted and observed breeding values for categorical traits (EP and STAY) and as the correlation between predicted and observed adjusted phenotypes divided by the square root of the estimated heritability for continuous traits (AFC and SC). The average ACC varied from low to moderate depending on the trait and model under consideration, ranging between 0.56 and 0.63 (AFC), 0.27 and 0.36 (SC), 0.57 and 0.67 (EP), and 0.52 and 0.62 (STAY). SVR provided slightly better accuracies than the parametric models for all traits, increasing the prediction accuracy for AFC to around 6.3 and 4.8% compared with GBLUP and BLASSO respectively. Likewise, there was an increase of 8.3% for SC, 4.5% for EP and 4.8% for STAY, comparing SVR with both GBLUP and BLASSO. In contrast, the RF and BRANN did not present competitive predictive ability compared with the parametric models. The results indicate that SVR is a suitable method for genome-enabled prediction of reproductive traits in Nellore cattle. Further, the optimal kernel bandwidth parameter in the SVR model was trait-dependent, thus, a fine-tuning for this hyper-parameter in the training phase is crucial.

Genomic reaction norm models exploiting genotype × environment interaction on sexual precocity indicator traits in Nellore cattle.

Brazilian beef cattle are raised predominantly on pasture in a wide range of environments. In this scenario, genotype by environment (G×E) interaction is an important source of phenotypic variation in the reproductive traits. Hence, the evaluation of G×E interactions for heifer's early pregnancy (HP) and scrotal circumference (SC) traits in Nellore cattle, belonging to three breeding programs, was carried out to determine the animal's sensitivity to the environmental conditions (EC). The dataset consisted of 85 874 records for HP and 151 553 records for SC, from which 1800 heifers and 3343 young bulls were genotyped with the BovineHD BeadChip. Genotypic information for 826 sires was also used in the analyses. EC levels were based on the contemporary group solutions for yearling body weight. Linear reaction norm models (RNM), using pedigree information (RNM_A) or pedigree and genomic information (RNM_H), were used to infer G×E interactions. Two validation schemes were used to assess the predictive ability, with the following training populations: (a) forward scheme-dataset was split based on year of birth from 2008 for HP and from 2011 for SC; and (b) environment-specific scheme-low EC (-3.0 and -1.5) and high EC (1.5 and 3.0). The inclusion of the H matrix in RNM increased the genetic variance of the intercept and slope by 18.55 and 23.00% on average respectively, and provided genetic parameter estimates that were more accurate than those considering pedigree only. The same trend was observed for heritability estimates, which were 0.28-0.56 for SC and 0.26-0.49 for HP, using RNM_H, and 0.26-0.52 for SC and 0.22-0.45 for HP, using RNM_A. The lowest correlation observed between unfavorable (-3.0) and favorable (3.0) EC levels were 0.30 for HP and -0.12 for SC, indicating the presence of G×E interaction. The G×E interaction effect implied differences in animals' genetic merit and re-ranking of animals on different environmental conditions. SNP marker-environment interaction was detected for Nellore sexual precocity indicator traits with changes in effect and variance across EC levels. The RNM_H captured G×E interaction effects better than RNM_A and improved the predictive ability by around 14.04% for SC and 21.31% for HP. Using the forward scheme increased the overall predictive ability for SC (20.55%) and HP (11.06%) compared with the environment-specific scheme. The results suggest that the inclusion of genomic information combined with the pedigree to assess the G×E interaction leads to more accurate variance components and genetic parameter estimates.

Genome scan for postmortem carcass traits in Nellore cattle.

Carcass traits measured after slaughter are economically relevant traits in beef cattle. In general, the slaughter house payment system is based on HCW. Ribeye area (REA) is associated with the amount of the meat in the carcass, and a minimum of backfat thickness (BFT) is necessary to protect the carcass during cooling. The aim of this study was to identify potential genomic regions harboring candidate genes affecting those traits in Nellore cattle. The data set used in the present study consisted of 1,756 Nellore males with phenotype records. A subset of 1,604 animals had both genotypic and phenotypic information. Genotypes were generated based on a panel with 777,962 SNPs from the Illumina Bovine HD chip. The SNP effects were calculated based on the genomic breeding values obtained by using the single-step GBLUP approach and a genomic matrix re-weighting procedure. The proportion of the variance explained by moving windows of 100 consecutive SNPs was used to assess potential genomic regions harboring genes with major effects on each trait. The top 10 non-overlapping SNP-windows explained 8.72%, 11.38%, and 9.31% of the genetic variance for REA, BFT, and HCW, respectively. These windows are located on chromosomes 5, 7, 8, 10, 12, 20, and 29 for REA; chromosomes 6, 8, 10, 13, 16, 17, 18, and 24 for BFT; and chromosomes 4, 6, 7, 8, 14, 16, 17, and 21 for HCW. For REA, there were identified genes ( and ) involved in the cell cycle biological process which affects many aspects of animal growth and development. The and genes, both from AA transporter family, was also associated with REA. The AA transporters are essential for cell growth and proliferation, acting as carriers of tissue nutrient supplies. Various genes identified for BFT (, , , , , and ) have been associated with lipid metabolism in different mammal species. One of the most promising genes identified for HCW was the . There is evidence, in the literature, that this gene is located in putative QTL affecting carcass weight in beef cattle. Our results showed several genomic regions containing plausible candidate genes that may be associated with carcass traits in Nellore cattle. Besides contributing to a better understanding of the genetic control of carcass traits, the identified genes can also be helpful for further functional genomic studies.

Genetic parameter estimates for carcass traits and visual scores including or not genomic information.

The objective of this study was to determine whether visual scores used as selection criteria in Nellore breeding programs are effective indicators of carcass traits measured after slaughter. Additionally, this study evaluated the effect of different structures of the relationship matrix ( and ) on the estimation of genetic parameters and on the prediction accuracy of breeding values. There were 13,524 animals for visual scores of conformation (CS), finishing precocity (FP), and muscling (MS) and 1,753, 1,747, and 1,564 for LM area (LMA), backfat thickness (BF), and HCW, respectively. Of these, 1,566 animals were genotyped using a high-density panel containing 777,962 SNP. Six analyses were performed using multitrait animal models, each including the 3 visual scores and 1 carcass trait. For the visual scores, the model included direct additive genetic and residual random effects and the fixed effects of contemporary group (defined by year of birth, management group at yearling, and farm) and the linear effect of age of animal at yearling. The same model was used for the carcass traits, replacing the effect of age of animal at yearling with the linear effect of age of animal at slaughter. The variance and covariance components were estimated by the REML method in analyses using the numerator relationship matrix () or combining the genomic and the numerator relationship matrices (). The heritability estimates for the visual scores obtained with the 2 methods were similar and of moderate magnitude (0.23-0.34), indicating that these traits should response to direct selection. The heritabilities for LMA, BF, and HCW were 0.13, 0.07, and 0.17, respectively, using matrix and 0.29, 0.16, and 0.23, respectively, using matrix . The genetic correlations between the visual scores and carcass traits were positive, and higher correlations were generally obtained when matrix was used. Considering the difficulties and cost of measuring carcass traits postmortem, visual scores of CS, FP, and MS could be used as selection criteria to improve HCW, BF, and LMA. The use of genomic information permitted the detection of greater additive genetic variability for LMA and BF. For HCW, the high magnitude of the genetic correlations with visual scores was probably sufficient to recover genetic variability. The methods provided similar breeding value accuracies, especially for the visual scores.

Associations between single nucleotide polymorphisms and carcass traits in Nellore cattle using high-density panels.

The objective of this study was to evaluate associations between single nucleotide polymorphism (SNP) markers and carcass traits measured postmortem in Nellore cattle. Records of loin eye area (LEA) and backfat thickness (BF) from 740 males and records of hot carcass weight (HCW) from 726 males were analyzed. All of the animals were genotyped using the BovineHD BeadChip. Association analyses were performed by the restricted maximum likelihood method that considered one SNP at a time. Significant SNPs were identified on chromosomes 2 and 6 for LEA and on chromosomes 7, 1, and 2 for BF. For HCW, associations with SNPs were found on chromosomes 13, 14, and 28, in addition to genome regions that were directly related to this trait, such as the EFCAB8 and VSTM2L genes, and to bone development (RHOU). Some SNPs were located in very close proximity to genes involved in basal metabolism (BLCAP, NNAT, CTNNBL1, TGM2, and LOC100296770) and the immune system (BPI).

Genetic association of growth traits with carcass and meat traits in Nellore cattle.

The aim of this study was to estimate genetic and phenotypic associations of growth traits with carcass and meat traits in Nellore cattle. Data from male and female animals were used for weaning weight (WW; N = 241,416), yearling weight (YW, N = 126,596), weight gain from weaning to yearling (GWY, N = 78,687), and yearling hip height (YHH, N = 90,720), respectively; 877 male animals were used for hot carcass weight (HCW) and 884 for longissimus muscle area (LMA), backfat thickness (BT), marbling score (MS), and shear force (SF). The variance components were estimated by the restricted maximum likelihood method using three-trait animal models that included WW. The model for WW included direct and maternal additive genetic, maternal permanent environmental, and residual effects as random effects; contemporary group as fixed effects; and age of dam at calving and age of animal as covariates (linear and quadratic effects). For the other traits, maternal effects and the effect of age of dam at calving were excluded from the model. Heritability ranged from 0.10 ± 0.12 (LMA) to 0.44 ± 0.007 (YW). Genetic correlations ranged from -0.40 ± 0.38 (WW x LMA) to 0.55 ± 0.10 (HCW x YW). Growth, carcass, and meat traits have sufficient genetic variability to be included as selection criteria in animal breeding programs.