Accuracy Of Estimated Breeding Values Research Articles

Comparative analysis of the heritability, estimated breeding value, and accuracy of carcass traits in Hanwoo from the Yeongnam Region, considering breeding area and climate effects, utilizing genomic information

Climate change is causing abnormal weather patterns worldwide. In Korea, local governments are characterized by various climatic conditions. Therefore, research integrating climate factors with Hanwoo is necessary. In this study, 562 Hanwoo were analyzed using Pedigree Best Linear Unbiased Prediction (PBLUP) and Genomic BLUP (GBLUP) models while considering climate factors such as temperature and humidity in the Yeongnam region. Carcass Weight (CWT) varied significantly in PBLUP depending on the model, while the heritability of Eye Muscle Area (EMA) and Back-Fat Thickness (BFT) decreased. However, the Marbling Score (MS) maintained high heritability across all models. GBLUP showed a slight decrease in the analysis model that considers breeding areas and climatic factors, yet still led to a high estimate of heritability for all traits. However, MS consistently had a heritability of 0.50 across all models. In PBLUP, the accuracy of the Estimated Breeding Value (EBV) increased by 0.15 for CWT (0.62-0.77) compared to Model 1. However, a decrease of 0.01–0.06 was observed for EMA (0.62–0.68) and BFT (0.53–0.58), while an increase of 0.01 was noted for MS (0.72–0.73). The accuracy of EBV using GBLUP was as follows: CWT (0.75-0.77), EMA (0.72-0.80), BFT (0.71-0.81), and MS (0.75). The accuracy of BFT significantly increased compared to PBLUP. For MS, it did not appear to be influenced by breeding area and climatic conditions. These results suggest the need for ongoing research on the relationship between Hanwoo carcass traits and climate effects, requiring additional test groups from various breeding areas.

A comparison of genetic and genomic breeding values in Saanen and Alpine goats

Nowadays, several countries are developing or adopting genomic selection in the dairy goat sector. The most used method to estimate breeding values is Single-Step Genomic Best Linear Unbiased Prediction (ssGBLUP) which offers several advantages in terms of computational process and accuracy of the estimated breeding values (EBVs). Saanen and Alpine are the predominant dairy goat breeds in Italy, and both have similar breeding programs where EBVs for productive traits are currently calculated using BLUP. This work describes the implementation of genomic selection for these two breeds in Italy, aligning with the selection practices already carried out in the international landscape. The available dataset included 3 611 genotyped animals, 11 470 lactation records, five traits (milk, protein and fat yields, and fat and protein percentages), and three-generation pedigrees. EBVs were estimated using BLUP, GBLUP, and ssGBLUP both with single and multiple trait approaches. The methods were compared in terms of correlation between EBVs and genetic trends. Results were also validated with the linear regression method excluding part of the phenotypic data. In both breeds, EBVs and GEBVs were strongly correlated and the trend of each trait was similar comparing the three methods. The average increase in accuracy across traits and methods amounted to +13 and +10% from BLUP to ssGBLUP for Alpine and Saanen breeds, respectively. Results indicated higher prediction accuracy and correlation for GBLUP and ssGBLUP compared to BLUP, implying that the use of genotypes increases the accuracy of EBVs, particularly in the absence of phenotypic data. Therefore, ssGBLUP is likely to be the most effective method to enhance genetic gain in Italian Saanen and Alpine goats.

Genetic parameters for carcass traits of progeny of beef bulls mated to dairy cows.

Beef × dairy crossbred cattle (n = 615) were used to evaluate the effect of preharvest indicator traits and genotypes on the accuracy of estimated breeding values (EBVs) of seedstock candidates for selection. Genotypes for 100,000 single nucleotide polymorphisms were provided by the American Simmental Association of purebred and crossbred seedstock animals (n = 2,632). Five hundred and ninety-five of the 615 beef × dairy cattle had carcass camera and ultrasound data. Phenotypes were not used for any of the seedstock animals even though some may have had performance and ultrasound data. We estimated the genomic relationship matrix among 3,247 animals including both phenotyped and unphenotyped animals. We computed genetic parameters among 37 traits using 666 bivariate restricted maximum likelihood analyses. The accuracy of EBV depends on heritability. For the sake of brevity, we report accuracy for marbling as a proxy for other traits with similar heritability. We focus on accuracy for marbling because marbling is the primary determinant of carcass value. We computed EBV for all 3,247 animals for marbling based on camera data postharvest using best linear unbiased prediction. We report evidence of overlap in causative genes among postharvest carcass traits; marbling, ribeye area, yield grade, fat thickness, and hot carcass weight (HCW) based on genetic correlations. Genetic correlations range from -0.73 to 0.89. Several live animal traits (frame size, body weight and ultrasound fat thickness and ribeye area) were genetically correlated with postharvest traits; including HCW, ribeye area, yield grade, fat thickness, and marbling. Genetic correlations between pre- and postharvest traits ranged from -0.53 to 0.95. Accuracy for marbling ranged from 0.64 to 0.80 for animals with marbling recorded, and from 0.09 to 0.60 for animals without marbling recorded. The accuracy of animals without phenotypes was related to the genomic relationship between animals with phenotype and those without. Live animal traits were useful for predicting economically important carcass traits based on genetic correlations. The accuracy of EBV for seedstock animals that were not phenotyped was low, but this is consistent with theory, and accuracy is expected to increase with the addition of genotypes and carcass data from beef × dairy animals.

Comparison between multiple-trait and random regression models for genetic evaluation of weight traits in Australian meat sheep.

Random regression (RR) models are recommended as an alternative to multiple-trait (MT) models for better capturing the variance-covariance structure over a trajectory and hence more accurate genetic evaluation of traits that are repeatedly measured and genetically change gradually over time. However, a limited number of studies have been done to empirically compare RR over a MT model to determine how much extra benefit could be achieved from one method over another. We compared the prediction accuracy of RR and MT models for growth traits of Australian meat sheep measured from 60 to 525d, using 102,579 weight records from 24,872 animals. Variance components and estimated breeding values (EBVs) estimated at specific ages were compared and validated with forward prediction. The accuracy of EBVs obtained from the MT model was 0.58, 0.51, 0.54, and 0.56 for weaning, postweaning, yearling, and hogget weight stages, respectively. RR model produced accuracy estimates of 0.56, 0.51, 0.54, and 0.54 for equivalent weight stages. Regression of adjusted phenotype on EBVs was very similar between the MT and the RR models (P > 0.05). Although the RR model did not significantly increase the accuracy of predicting future progeny performance, there are other benefits of the model such as no limit to the number of records per animal, estimation of EBVs for early and late growth, no need for age correction. Therefore, RR can be considered a more flexible method for the genetic evaluation of Australian sheep for early and late growth, and no need for age correction.

PSVII-3 The Impact of Genotyping Strategies on Accuracy of Genomic Prediction in Sheep Populations: Preliminary Results of a Simulation Study

Abstract Genomic predictions provide more accurate estimated breeding values (EBV) in younger animals, enabling greater rates of genetic gains over time. However, genomic selection in U.S. sheep is still incipient due to the higher genotyping cost relative to animal cost. Our objective was to compare genotyping strategies differing in the proportion of animals genotyped and the extent of pedigree error on prediction accuracy. These genetic evaluations were based on single-step Genomic Best Linear Unbiased Prediction. We used the AlphaSimR package to simulate a composite sheep population mimicking the Katahdin breeds’ formation. We then simulated four further generations based on the structure of Katahdin flocks engaged in the National Sheep Improvement Program. Thirty-five flocks were simulated with an effective population size of 170. Flock size ranged from 31 to 680 animals and, by design, was skewed towards smaller flocks. Average ram to ewe ratio, mortality rate per generation, and number of lambs born per ewe lambing was 1:16, 10%, and 1.8 lambs, respectively. We simulated a 50k SNP panel for a moderately heritable (0.30) trait. Flocks had low genetic connectedness. Selection criteria differed among flocks: 3 selected randomly, 10 selected based on highest EBV, and 22 selected based on highest phenotypes. All combinations of 5, 40, or 100% of males or females genotyped, 5 or 20% of animals missing pedigree, and 5 or 20% of sires misidentified were compared. We also considered a scenario with no pedigree errors (i.e., no missing pedigree or misidentified sires). Animals with the best phenotype were chosen for genotyping. The first three generations were used as the reference population (7,317 animals) to validate the 4th generation (2,055 animals). Each scenario was replicated 5 times. A linear model was used to estimate differences in prediction accuracy between scenarios, tested with Tukey’s test (alpha 0.05). For the scenario without pedigree errors, prediction accuracy ranged from 0.50 ± 0.06 for 5% of males and females genotyped to 0.75 ± 0.03 for 100% of animals genotyped (Figure 1). For scenarios with 5% pedigree errors, prediction accuracy ranged from 0.33 ± 0.02 for 5% of males and females genotyped to 0.71 ± 0.02 for 100% of animals genotyped. For the scenario with 20% pedigree errors, prediction accuracy ranged from 0.26 ± 0.06 for 5% of males and females genotyped to 0.67 ± 0.03 for 100% of animals genotyped. Across pedigree error scenarios, trends were similar. Importantly, genotyping 5% of one sex and 100% of the other resulted in similar accuracies to genotyping 100% of animals. In scenarios with few animals genotyped, accuracies were higher when errors in pedigree were avoided. These preliminary results reinforce the importance of genomics in correcting pedigree errors, with high levels of prediction accuracy when at least most animals of one sex are genotyped.

Comparison of accuracy of breeding value for cow from three methods in Hanwoo (Korean cattle) population.

In Korea, Korea Proven Bulls (KPN) program has been well-developed. Breeding and evaluation of cows are also an essential factor to increase earnings and genetic gain. This study aimed to evaluate the accuracy of cow breeding value by using three methods (pedigree index [PI], pedigree-based best linear unbiased prediction [PBLUP], and genomic-BLUP [GBLUP]). The reference population (n = 16,971) was used to estimate breeding values for 481 females as a test population. The accuracy of GBLUP was 0.63, 0.66, 0.62 and 0.63 for carcass weight (CWT), eye muscle area (EMA), back-fat thickness (BFT), and marbling score (MS), respectively. As for the PBLUP method, accuracy of prediction was 0.43 for CWT, 0.45 for EMA, 0.43 for MS, and 0.44 for BFT. Accuracy of PI method was the lowest (0.28 to 0.29 for carcass traits). The increase by approximate 20% in accuracy of GBLUP method than other methods could be because genomic information may explain Mendelian sampling error that pedigree information cannot detect. Bias can cause reducing accuracy of estimated breeding value (EBV) for selected animals. Regression coefficient between true breeding value (TBV) and GBLUP EBV, PBLUP EBV, and PI EBV were 0.78, 0.625, and 0.35, respectively for CWT. This showed that genomic EBV (GEBV) is less biased than PBLUP and PI EBV in this study. In addition, number of effective chromosome segments (Me) statistic that indicates the independent loci is one of the important factors affecting the accuracy of BLUP. The correlation between Me and the accuracy of GBLUP is related to the genetic relationship between reference and test population. The correlations between Me and accuracy were -0.74 in CWT, -0.75 in EMA, -0.73 in MS, and -0.75 in BF, which were strongly negative. These results proved that the estimation of genetic ability using genomic data is the most effective, and the smaller the Me, the higher the accuracy of EBV.

Comparison on genomic prediction using pedigree BLUP and single step GBLUP through the Hanwoo full-sib family.

When evaluating individuals with the same parent and no phenotype by pedigree best linear unbiased prediction (BLUP), it is difficult to explain carcass grade difference and select individuals because they have the same value in pedigree BLUP (PBLUP). However, single step GBLUP (ssGBLUP), which can estimate the breeding value suitable for the individual by adding genotype, is more accurate than the existing method. The breeding value and accuracy were estimated with pedigree BLUP and ssGBLUP using pedigree and genotype of 408 Hanwoo cattle from 16 families with the same parent among siblings produced by fertilized egg transplantation. A total of 14,225 Hanwoo cattle with pedigree, genotype and phenotype were used as the reference population. PBLUP obtained estimated breeding value (EBV) using the pedigree of the test and reference populations, and ssGBLUP obtained genomic EBV (GEBV) after constructing and H-matrix by integrating the pedigree and genotype of the test and reference populations. For all traits, the accuracy of GEBV using ssGBLUP is 0.18 to 0.20 higher than the accuracy of EBV obtained with PBLUP. Comparison of EBV and GEBV of individuals without phenotype, since the value of EBV is estimated based on expected values of alleles passed down from common ancestors. It does not take Mendelian sampling into consideration, so the EBV of all individuals within the same family is estimated to be the same value. However, GEBV makes estimating true kinship coefficient based on different genotypes of individuals possible, so GEBV that corresponds to each individual is estimated rather than a uniform GEBV for each individual. Since Hanwoo cows bred through embryo transfer have a high possibility of having the same parent, if ssGBLUP after adding genotype is used, estimating true kinship coefficient corresponding to each individual becomes possible, allowing for more accurate estimation of breeding value.

Strategies for accommodating gene-edited sires and their descendants in genetic evaluations.

Gene editing has the potential to expedite the rate of genetic gain for complex traits. However, changing nucleotides (i.e., QTN) in the genome can affect the additive genetic relationship among individuals and, consequently, impact genetic evaluations. Therefore, the objectives of this study were to estimate the impact of including gene-edited individuals in the genetic evaluation and investigate modeling strategies to mitigate potential errors. For that, a beef cattle population was simulated for nine generations (N = 13,100). Gene-edited sires (1, 25, or 50) were introduced in generation 8. The number of edited QTN was 1, 3, or 13. Genetic evaluations were performed using pedigree, genomic data, or a combination of both. Relationships were weighted based on the effect of the edited QTN. Comparisons were made using the accuracy, average absolute bias, and dispersion of the estimated breeding values (EBV). In general, the EBV of the first generation of progeny of gene-edited sires were associated with greater average absolute bias and overdispersion than the EBV of the progeny of non-gene-edited sires (P ≤ 0.001). Weighting the relationship matrices increased (P ≤ 0.001) the accuracy of EBV when the gene-edited sires were introduced by 3% and decreased (P ≤ 0.001) the average absolute bias and dispersion for the progeny of gene-edited sires. For the second generation of descendants of gene-edited sires, the absolute bias increased as the number of edited alleles increased; however, the rate of increase in absolute bias was 0.007 for each allele edited when the relationship matrices were weighted compared with 0.10 when the relationship matrices were not weighted. Overall, when gene-edited sires are included in genetic evaluations, error is introduced in the EBV, such that the EBV of progeny of gene-edited sires are underestimated. Hence, the progeny of gene-edited sires would be less likely to be selected to be parents of the next generation than what was expected based on their true genetic merit. Therefore, modeling strategies such as weighting the relationship matrices are essential to avoid incorrect selection decisions if animals that have been edited for QTN underlying complex traits are introduced into genetic evaluations.

Estimation of breeding values using different densities of SNP to inform kinship in broiler chickens

Traditionally, breeding values are estimated based on phenotypic and pedigree information using the numerator relationship (A) matrix. With the availability of genomic information, genome-wide markers can be included in the estimation of breeding values through genomic kinship. Genomic information allows a more accurate estimation of breeding values compared to the breeding values estimated by phenotypic and pedigree information. However, the density of genomic information used can impact the cost of implementation. The aim of this study was to compare the rank, accuracy, and bias of estimated breeding values (EBV) for organs [heart (HRT), liver (LIV), gizzard (GIZ), lungs (LUN)] and carcass [breast (BRST), drumstick (DRM) and thigh (THG)] weight traits in a broiler population using pedigree-based BLUP (PBLUP) and single-step genomic BLUP (ssGBLUP) methods using various densities of SNP and variants imputed from whole-genome sequence (WGS). For both PBLUP and ssGBLUP, heritability estimates varied from low (LUN) to high (HRT, LIV, GIZ, BRST, DRM and THG). Regression coefficient values of EBV on genomic estimated breeding values (GEBV) were similar for both the high density (HD) and WGS sets of SNPs, ranging from 0.87 to 0.99 across scenarios. Results show no benefit of using WGS data compared to HD array data using an unweighted ssGBLUP. Our results suggest that 10% of the content of the HD array can yield unbiased and accurate EBV.

Impact and utility of shallow pedigree using single-step genomic BLUP for prediction of unbiased genomic breeding values.

In unstructured dairy programs, pedigree is usually shallow, which leads to biased prediction of breeding values using best linear unbiased prediction (BLUP). The objective of this study was to come out with a genomic prediction strategy that can utilize shallow pedigree information and predict unbiased and more accurate GEBV for sex-limited traits in a small population using single-step GBLUP (ssGBLUP). The data and models for a population under selection were simulated. Out of current 10 generations, 10th generation with 1000 candidates served as validation population. For the complete pedigree scenario, pedigree (P)BLUP estimated breeding values (EBV) were unbiased with accuracy (r) of 0.35 ± 0.02 and 0.26 ± 0.01 for 0.3 and 0.1 h2 scenario, respectively. For the shallow pedigree, biased prediction of breeding values and low accuracies were obtained with linear decline in the accuracy of EBV for removal of information on more distant pedigree. Accuracy and bias (ρ) for scenario with removing 4 distant generations from pedigree were 0.30 ± 0.02 and 0.55 ± 0.03, respectively, in moderate h2 scenario. Use of Genomic (G)BLUP, especially with "extreme phenotypic contrast selective genotyping," (TB) resulted in higher accuracy for a small reference of females; however, GEBV were highly biased. We observed that ssGBLUPF, where the numerator relationship matrix is corrected for inbreeding, resulted in more accurate and unbiased estimates of GEBV across shallow pedigree scenario, with TB all female reference (missing 4 distant generations: r = 0.50 ± 0.02; ρ = 0.96 ± 0.02). We recommend use of ssGBLUPF with two tailed selectively genotyped all female reference in shallow pedigree scenarios, to obtain unbiased and accurate GEBV for sex-limited traits, when resources are limited.

PSVI-21 Including Gene-Edited Individuals in Genetic Evaluations can Bias the Estimated Breeding Values of Their Progeny

Abstract Gene-editing offers the potential to increase genetic gain for complex traits, but practical issues for routine genetic evaluations need to be addressed to ensure kinship-based predictions are not biased. The objective of this study was to quantify differences in estimated breeding values (EBV) using a 8-generation simulated beef cattle population with progeny of gene-edited (GE) sires. After generating haplotypes, 99 quantitative trait nucleotides (QTN) and 6,000 single nucleotide polymorphisms (SNP) were distributed across 3 chromosomes. A single trait (h2=0.4) was simulated. After 8 generations, GE sires (n=1, 25, or 50) were introduced. The number of QTN edited (% additive variation) was 1 (2%), 3 (5%), or 13 (20%). All scenarios were replicated 15 times. Genetic evaluations were performed using pedigree (A), genomic (G), or combined (H) kinship matrices. Relationships were also weighted (w) based on the proportion of genetic variance explained by the edited QTN. To create Aw, weights were based on the expected difference between G and Gw. Hw was formed using Aw and Gw. Scenarios were compared using accuracy (correlation of true BV (TBV) and EBV), average absolute bias, and the slope (b1) of the regression of TBV on EBV. Accuracy of EBV between GE vs. non-GE sired progeny did not differ. EBV of GE sired progeny were under-dispersed (b1 from 0.61 to 0.92) and associated with greater average absolute bias (0.8 to 3.8), and these issues increased in severity with the number of edited QTN. Weighting kinship reduced the average absolute bias (from 2.7 to 0.9 with 13 edited QTN) and the dispersion of EBV (from 0.7 to 0.9 with 13 edited QTN). The same pattern was observed across all kinship matrices. Results indicate that including GE individuals in genetic evaluations could lead to biased EBV, but weighting kinship could partially mitigate this concern.

International single-step SNPBLUP beef cattle evaluations for Limousin weaning weight

BackgroundCompared to national evaluations, international collaboration projects further improve accuracies of estimated breeding values (EBV) by building larger reference populations or performing a joint evaluation using data (or proxy of them) from different countries. Genomic selection is increasingly adopted in beef cattle, but, to date, the benefits of including genomic information in international evaluations have not been explored. Our objective was to develop an international beef cattle single-step genomic evaluation and investigate its impact on the accuracy and bias of genomic evaluations compared to current pedigree-based evaluations.MethodsWeaning weight records were available for 331,593 animals from seven European countries. The pedigree included 519,740 animals. After imputation and quality control, 17,607 genotypes at a density of 57,899 single nucleotide polymorphisms (SNPs) from four countries were available. We implemented two international scenarios where countries were modelled as different correlated traits: an international genomic single-step SNP best linear unbiased prediction (SNPBLUP) evaluation (ssSNPBLUPINT) and an international pedigree-based BLUP evaluation (PBLUPINT). Two national scenarios were implemented for pedigree and genomic evaluations using only nationally submitted phenotypes and genotypes. Accuracies, level and dispersion bias of EBV of animals born from 2014 onwards, and increases in population accuracies were estimated using the linear regression method.ResultsOn average across countries, 39 and 17% of sires and maternal-grand-sires with recorded (grand-)offspring across two countries were genotyped. ssSNPBLUPINT showed the highest accuracies of EBV and, compared to PBLUPINT, led to increases in population accuracy of 13.7% for direct EBV, and 25.8% for maternal EBV, on average across countries. Increases in population accuracies when moving from national scenarios to ssSNPBLUPINT were observed for all countries. Overall, ssSNPBLUPINT level and dispersion bias remained similar or slightly reduced compared to PBLUPINT and national scenarios.ConclusionsInternational single-step SNPBLUP evaluations are feasible and lead to higher population accuracies for both large and small countries compared to current international pedigree-based evaluations and national evaluations. These results are likely related to the larger multi-country reference population and the inclusion of phenotypes from relatives recorded in other countries via single-step international evaluations. The proposed international single-step approach can be applied to other traits and breeds.

Consequences of specialized breeding in the Swedish Warmblood horse population.

In many European warmblood studbooks, clear specialization toward either jumping or dressage horses is evident. The Swedish Warmblood (SWB) is also undergoing such specialization, creating a possible need for separate breeding programs and a discipline-specific Young Horse Test (YHT). This study investigated how far specialization of the SWB breed has proceeded and the potential consequences. Individuals in a population of 122,054 SWB horses born between 1980 and 2020 were categorized according to pedigree as jumping (J), dressage (D), allround (AR), or thoroughbred (Th). Data on 8,713 J horses and 6,477 D horses assessed for eight traits in YHT 1999-2020 were used to estimate genetic parameters within and between J and D horses and between different periods. Future scenarios in which young horses are assessed for either jumping or dressage traits at YHT were also analyzed. More than 80% of horses born in 1980-1985 were found to be AR horses, while 92% of horses born in 2016-2020 belonged to a specialized category. The average relationship within J or D category was found to increase during the past decade, whereas the relationship between these categories decreased. Heritability estimates for gait traits were 0.42-0.56 for D horses and 0.25-0.38 for J horses. For jumping traits, heritability estimates were 0.17-0.26 for J horses and 0.10-0.18 for D horses. Genetic correlations between corresponding traits assessed in J and D horses were within the range 0.48-0.81, with a tendency to be lower in the late study period. In the future scenarios, heritability and genetic variance both decreased for traits that were not assessed in all horses, indicating that estimation of breeding value and genetic progress for these traits could be affected by a specialized YHT. However, ranking of sires based on estimated breeding values (EBVs) and accuracy of EBVs was only slightly altered for discipline-specific traits. With continued specialization in SWB, specialization of the YHT should thus be considered.

SNP arrays evaluation as tools in genetic improvement in Corriedale sheep in Uruguay

One control strategy for gastrointestinal nematodes (GIN) is genetic selection. This study´s objective was to compare eggs per gram of feces (FEC) and fiber diameter (FD) estimated breeding values (EBV) and genomic EBV (GEBV) in Corriedale breed. Analysis included 19547 lambs with data, and 454, 711 and 383 genotypes from 170, 507 and 50K SNP chips, respectively. A univariate animal model was used for EBV and GEBV estimation, which included contemporary group, type of birth and dam age as fixed effects, and age at recording as covariate. Differential weights (α) were considered in the genomic relationship matrix (G), and the best fit models were identified using Akaike´s Information Criterion (AIC), which were later used for GEBV and accuracies estimation. The use of α only impacted on low density SNP chips. No differences were observed in mean accuracies for the whole population. However, in the genotyped subgroup accuracies increased by 2% with the 170 SNP chip (α=0.25), and by 5% (α=0.5) and 14% (α=0.75) with the 507 SNP chip. No differences were observed in FD EBV and GEBV mean accuracies. These results show that it is possible to increase GEBV accuracies despite the use of low-density chips.

43 Single and Multiple-breed Genomic Predictions for Conformation Traits of Canadian Dairy Goats

Abstract The use of multiple-breed models can increase the accuracy of estimated breeding values (EBV) when few phenotypes are available for a trait. However, pooling breeds is not always beneficial for genomic evaluations due to the low consistency of gametic phase between individual breeds. The objective of this study was to compare the expected gain in accuracy of single-step genomic breeding values (GEBV) for conformation traits of Canadian Alpine and Saanen goats predicted using single and multiple-breed models. The traits considered were body capacity, dairy character, feet and legs, fore udder, general appearance, rear udder, suspensory ligament, and teats, all recorded by trained classifiers, using a 1 to 9 scale. The full datasets included a total of 7,500 phenotypes for each trait (5,158 Alpine and 2,342 Saanen) and 1,707 50K genotypes (833 Alpine, 874 Saanen). Standard errors of prediction (SEP) were obtained for EBV and GEBV predicted using single-trait animal models on full or validation datasets. Breed difference was accounted for as a fixed effect in the multiple-breed models. Average theoretical accuracies were calculated from the SEP. For Saanen, with fewer records, expected accuracies of EBV and GEBV for the validation animals (selection candidates) were consistently higher for the multiple-breed models. Trait specific gains in theoretical accuracy of GEBV relative to EBV for the selection candidates ranged from 30 to 48% for Alpine and 41 to 61% for Saanen. Averaged across all traits, GEBV predicted from the full dataset were 32 to 38% more accurate than EBV for genotyped animals and the largest gains were found for does without conformation records (49 to 55%) and bucks without daughter records (56 to 82%). Overall, the implementation of genomic selection would substantially increase selection accuracy for young breeding candidates and, consequently, the rate of genetic improvement for conformation traits in Canadian dairy goats.

Short communication: investigation of the feasibility of genomic selection in Icelandic Cattle.

Icelandic Cattle is a local dairy cattle breed in Iceland. With about 26,000 breeding females, it is by far the largest among the indigenous Nordic cattle breeds. The objective of this study was to investigate the feasibility of genomic selection in Icelandic Cattle. Pedigree-based best linear unbiased prediction (PBLUP) and single-step genomic best linear unbiased prediction (ssGBLUP) were compared. Accuracy, bias, and dispersion of estimated breeding values (EBV) for milk yield (MY), fat yield (FY), protein yield (PY), and somatic cell score (SCS) were estimated in a cross validation-based design. Accuracy (r^) was estimated by the correlation between EBV and corrected phenotype in a validation set. The accuracy (r^) of predictions using ssGBLUP increased by 13, 23, 19, and 20 percentage points for MY, FY, PY, and SCS for genotyped animals, compared with PBLUP. The accuracy of nongenotyped animals was not improved for MY and PY, but increased by 0.9 and 3.5 percentage points for FY and SCS. We used the linear regression (LR) method to quantify relative improvements in accuracy, bias (Δ^), and dispersion (b^) of EBV. Using the LR method, the relative improvements in accuracy of validation from PBLUP to ssGBLUP were 43%, 60%, 50%, and 48% for genotyped animals for MY, FY, PY, and SCS. Single-step GBLUP EBV were less underestimated (Δ^), and less overdispersed (b^) than PBLUP EBV for FY and PY. Pedigree-based BLUP EBV were close to unbiased for MY and SCS. Single-step GBLUP underestimated MY EBV but overestimated SCS EBV. Based on the average accuracy of 0.45 for ssGBLUP EBV obtained in this study, selection intensities according to the breeding scheme of Icelandic Cattle, and assuming a generation interval of 2.0 yr for sires of bulls, sires of dams and dams of bulls, genetic gain in Icelandic Cattle could be increased by about 50% relative to the current breeding scheme.

38 Using pooled data for single-step genomic prediction: Impact of within-pool variance and size

Abstract Economically relevant traits (ERT) are routinely collected within commercial segments of the beef industry but are rarely included in genetic evaluations because of unknown pedigrees. Individual relationships could be resurrected with genomics, which would be costly; pooling DNA and phenotypic data provides a cost-effective solution. A simulated beef cattle population consisting of 15 generations was genotyped with approximately 50k markers (841 quantitative trait loci were located across the genome) and phenotyped for a moderately heritable trait. Individuals from generation 15 were included in pools (observed genotype and phenotype were mean values of a group). Estimated breeding values (EBV) were generated from a single-step GBLUP model. The effects of pooling strategy (random and minimizing or uniformly maximizing phenotypic variation), pool size (1, 2, 10, 20, 50, 100, or no data from generation 15), and generational gaps of genotyping on EBV accuracy (correlation of EBV with true breeding values) were quantified. Greatest EBV accuracies of sires and dams were observed when no gap between genotyped parents and pooled offspring occurred. The EBV accuracies resulting from pools were greater than no data from generation 15 regardless of sire or dam genotyping. Minimizing phenotypic variation increased EBV accuracy by 8% and 9% over random pooling and uniformly maximizing phenotypic variation, respectively. Pool size of 2 was the only scenario that did not significantly decrease EBV accuracy compared to individual data when pools were formed randomly or by uniformly maximizing phenotypic variation (P > 0.05). Pool sizes of 2, 10, 20, or 50 did not generally lead to EBV accuracies that were statistically different than individual data when pools were constructed to minimize phenotypic variation (P > 0.05). Pooled genotyping to garner commercial-level phenotypes for genetic evaluations seems plausible, although differences exist depending on pool size and pool formation strategy. The USDA is an equal opportunity employer.

PSXII-37 Validation of single-step GBLUP genomic predictions from threshold models using the linear regression method: an application in chicken mortality

Abstract The objective of this study was to determine whether the linear regression (LR) method could be used to validate genomic threshold models used for analysis of categorical data. Statistics for the LR method were computed from estimated breeding values (EBVs) using the whole and truncated datasets. The method was tested using simulated and real chicken datasets. The simulated dataset included ten generations of 4,500 birds each; genotypes were available for the last three generations. Each animal was assigned a continuous trait, which was converted to a binary score assuming an incidence of failure of 7%. The real dataset included the survival status of 186,596 broilers (mortality rate equal to 7.2%) and genotypes of 18,047 birds. Both datasets were analyzed using Best Linear Unbiased Predictor (BLUP) or single-step GBLUP (ssGBLUP). The whole dataset included all phenotypes available, whereas in the partial dataset, phenotypes of the most recent generation were removed. In the simulated dataset, the accuracies based on the LR formulas were 0.45 for BLUP and 0.76 for ssGBLUP, whereas the correlations between true breeding values and EBVs were 0.37 and 0.65, respectively. The gain in accuracy by adding genomic information was overestimated by 0.09 when using the LR method compared to the true increase in accuracy. However, when the estimated ratio between the additive variance computed based on pedigree only and on pedigree and genomic information was considered, the difference between true and estimated gain was less than 0.02. Accuracies of BLUP and ssGBLUP with the real dataset were 0.41 and 0.47, respectively. Smaller improvements in accuracy when using ssGBLUP with the real dataset were due to population structure and lower heritability. The LR method is a useful tool for estimating improvements in accuracy of EBVs due to the inclusion of genomic information when traditional validation methods are not applicable.

384 Genetic and Genomic Analysis in Livestock with Increasing Datasets

Abstract In animal breeding and genetics, statistical methods have been used for several decades to identify animals with the best genetic potential. One of the foundations for computing accurate estimated breeding values (EBV) is the amount of data that is used in the evaluation system — as the more data points one animal has, the more accurate its EBV is going to be. However, the animal breeding and genetics field periodically faces a big data paradox, where efficient methods have to be developed to handle the amount of data collected over time, given the computing capacity becomes the limiting factor. For instance, running genetic evaluations based on phenotypes and pedigree for a million animals was impossible in 1970. Methods and algorithms evolved to a point where using data for millions of animals was not a problem, until genomic information became available. After the development of single nucleotide polymorphism (SNP) chips for livestock in 2008, genomic information started being used in addition to phenotypes and pedigree to further improve accuracy of EBV. However, each animal is genotyped for around 50,000 SNP, which makes this data dense and difficult to work with. Over 790,000 Angus and 3.4 million Holstein cattle have been genotyped in the US as of April 2020. As the amount of new data considerably increases every week, most of the genomic evaluations are done on a weekly basis. Given that data have to be processed, the computation of EBV cannot take more than four days, which can be challenging depending on the model. In this talk we will discuss the challenges and solutions for successful genomic evaluations in large livestock populations. Finally, perspectives on the use of whole-genome sequence data and high-throughput phenotypes in genomic analysis will be summarized.

Validation of single-step GBLUP genomic predictions from threshold models using the linear regression method: An application in chicken mortality.

The objective of this study was to determine whether the linear regression (LR) method could be used to validate genomic threshold models. Statistics for the LR method were computed from estimated breeding values (EBVs) using the whole and truncated data sets with variances from the reference and validation populations. The method was tested using simulated and real chicken data sets. The simulated data set included 10 generations of 4,500 birds each; genotypes were available for the last three generations. Each animal was assigned a continuous trait, which was converted to a binary score assuming an incidence of failure of 7%. The real data set included the survival status of 186,596 broilers (mortality rate equal to 7.2%) and genotypes of 18,047 birds. Both data sets were analysed using best linear unbiased predictor (BLUP) or single‐step GBLUP (ssGBLUP). The whole data set included all phenotypes available, whereas in the partial data set, phenotypes of the most recent generation were removed. In the simulated data set, the accuracies based on the LR formulas were 0.45 for BLUP and 0.76 for ssGBLUP, whereas the correlations between true breeding values and EBVs (i.e. true accuracies) were 0.37 and 0.65, respectively. The gain in accuracy by adding genomic information was overestimated by 0.09 when using the LR method compared to the true increase in accuracy. However, when the estimated ratio between the additive variance computed based on pedigree only and on pedigree and genomic information was considered, the difference between true and estimated gain was <0.02. Accuracies of BLUP and ssGBLUP with the real data set were 0.41 and 0.47, respectively. This small improvement in accuracy when using ssGBLUP with the real data set was due to population structure and lower heritability. The LR method is a useful tool for estimating improvements in accuracy of EBVs due to the inclusion of genomic information when traditional validation methods as k‐fold validation and predictive ability are not applicable.