Bias Of Genomic Predictions Research Articles

Genomic evaluation of body weight traits in a F2 mixture of commercial broiler and native chicken

Abstract Genetic improvement of body weight (BW) traits has received major consideration in the poultry industry due to their economic and environmental implications. With the rapid implementation of genomic selection (GS) in the poultry industry and a decrease in the cost of genotyping, genomic prediction (GP) is a feasible way to increase productivity. Moreover, a pre-selection of SNPs could represent a reasonable option to speed up GP. We used 312 F2 broiler chicken genotyped with 60K Illumina Beadchip to investigate the effect of reduced SNP densities on accuracy and bias of prediction using single-step genomic BLUP (ssGBLUP) for BW at 2-4 weeks of age (488 chickens). To investigate the effect of reduced SNP densities by varying minor allele frequency (MAF), SNPs were grouped into five subgroups with MAF of 0.05-0.1, 0.1-0.2, 0.2-0.3, 0.3-0.4 and 0.4-0.5. The accuracy and bias of genomic predictions from different MAF bins were compared to that using a standard array of 60k SNP genotypes and the traditional BLUP method. Our study showed that using a subset of common SNPs genotypes may increase accuracy of genomic predictions compared to using all SNPs, specifically in the studied F2 population with a limited number of genotyped/phenotyped individuals.

Impact of accounting for allele dosage on genomic prediction in a simulated autopolyploid sturgeon population

Genomic selection is an efficient approach to get shorter breeding cycles in selection programs and greater genetic gains with the selection of superior individuals. Despite advances in genotyping techniques, genetic studies for polyploid species are limited to a rough approximation due to lacking of proper method and having to use the method of diploid species. The major challenge of genetic evaluation is to distinguish the different types of heterozygotes presented in polyploid populations. The main objective of this study was to investigate the impact of accounting for allele dosage on the accuracy and bias of genomic prediction in autopolyploid sturgeons with different ploidy levels. We simulated an autopolyploid sturgeon population of tetraploid and octoploid. Three methods, including BLUP, genomic BLUP (GBLUP) and reproducing kernel Hilbert space (RKHS), were implemented to investigate their prediction abilities. Moreover, the impact of double reduction on prediction accuracy and the impact of the reference population composition on genomic prediction were also surveyed. The results showed that there was no difference in prediction accuracy between pseudo diploid and autopolyploid with different double reductions by BLUP method. However, considering the allele dosage in genomic prediction models promoted approximately 5.1% and 13.0% genomic prediction accuracy in tetraploid and octoploid respectively, compared with using pseudo diploid allele dosage. GBLUP and RKHS models showed superiority over BLUP model, and RKHS performed equal to GBLUP in all scenarios. In addition, reference population only comprising top individuals led to low accuracy and serious bias of genomic prediction. Our results can be applied to optimize genomic prediction for other species, especially polyploids.

Improving Genomic Prediction of Crossbred and Purebred Dairy Cattle.

This study assessed the accuracy and bias of genomic prediction (GP) in purebred Holstein (H) and Jersey (J) as well as crossbred (H and J) validation cows using different reference sets and prediction strategies. The reference sets were made up of different combinations of 36,695 H and J purebreds and crossbreds. Additionally, the effect of using different sets of marker genotypes on GP was studied (conventional panel: 50k, custom panel enriched with, or close to, causal mutations: XT_50k, and conventional high-density with a limited custom set: pruned HDnGBS). We also compared the use of genomic best linear unbiased prediction (GBLUP) and Bayesian (emBayesR) models, and the traits tested were milk, fat, and protein yields. On average, by including crossbred cows in the reference population, the prediction accuracies increased by 0.01–0.08 and were less biased (regression coefficient closer to 1 by 0.02–0.16), and the benefit was greater for crossbreds compared to purebreds. The accuracy of prediction increased by 0.02 using XT_50k compared to 50k genotypes without affecting the bias. Although using pruned HDnGBS instead of 50k also increased the prediction accuracy by about 0.02, it increased the bias for purebred predictions in emBayesR models. Generally, emBayesR outperformed GBLUP for prediction accuracy when using 50k or pruned HDnGBS genotypes, but the benefits diminished with XT_50k genotypes. Crossbred predictions derived from a joint pure H and J reference were similar in accuracy to crossbred predictions derived from the two separate purebred reference sets and combined proportional to breed composition. However, the latter approach was less biased by 0.13. Most interestingly, using an equalized breed reference instead of an H-dominated reference, on average, reduced the bias of prediction by 0.16–0.19 and increased the accuracy by 0.04 for crossbred and J cows, with a little change in the H accuracy. In conclusion, we observed improved genomic predictions for both crossbreds and purebreds by equalizing breed contributions in a mixed breed reference that included crossbred cows. Furthermore, we demonstrate, that compared to the conventional 50k or high-density panels, our customized set of 50k sequence markers improved or matched the prediction accuracy and reduced bias with both GBLUP and Bayesian models.

Applying an association weight matrix in weighted genomic prediction of boar taint compounds.

Biological information regarding markers and gene association may be used to attribute different weights for single nucleotide polymorphism (SNP) in genome-wide selection. Therefore, we aimed to evaluate the predictive ability and the bias of genomic prediction using models that allow SNP weighting in the genomic relationship matrix (G) building, with and without incorporating biological information to obtain the weights. Firstly, we performed a genome-wide association studies (GWAS) in data set containing single- (SL) or a multi-line (ML) pig population for androstenone, skatole and indole levels. Secondly, 1%, 2%, 5%, 10%, 30% and 50% of the markers explaining the highest proportions of the genetic variance for each trait were selected to build gene networks through the association weight matrix (AWM) approach. The number of edges in the network was computed and used to derive weights for G (AWM-WssGBLUP). The single-step GBLUP (ssGBLUP) and weighted ssGBLUP (WssGBLUP) were used as standard scenarios. All scenarios presented predictive abilities different from zero; however, the great overlap in their confidences interval suggests no differences among scenarios. Most of scenarios of based on AWM provide overestimations for skatole in both SL and ML populations. On the other hand, the skatole and indole prediction were no biased in the ssGBLUP (S1) in both SL and ML populations. Most of scenarios based on AWM provide no biased predictions for indole in both SL and ML populations. In summary, using biological information through AWM matrix and gene networks to derive weights for genomic prediction resulted in no increase in predictive ability for boar taint compounds. In addition, this approach increased the number of analyses steps. Thus, we can conclude that ssGBLUP is most appropriate for the analysis of boar taint compounds in comparison with the weighted strategies used in the present work.

Bias in genomic predictions by mating practices for linear type traits in a large-scale genomic evaluation

The objective of this study was to clarify how bias in genomic predictions is created by investigating a relationship among selection intensity, a change in heritability (Δh2), and assortative mating (ASM). A change in heritability, resulting from selection, reflects the impact that the Bulmer effect has on the reduction in between-family variation, whereas assortative mating impacts the within-family variance or Mendelian sampling variation. A partial data set up to 2014, including 841K genotyped animals, was used to calculate genomic predictions with a single-step genomic model for 18 linear type traits in US Holsteins. A full data set up to 2018, including 2.3 million genotyped animals, was used to calculate benchmark genomic predictions. Inbreeding and unknown parent groups for missing parents of animals were included in the model. Genomic evaluation was performed using 2 different genetic parameters: those estimated 14 yr ago, which have been used in the national genetic evaluation for linear type traits in the United States, and those newly estimated with recent records from 2015 to 2018 and those corresponding pedigrees. Genetic trends for 18 type traits were estimated for bulls with daughters and cows with phenotypes in 2018. Based on selection intensity and mating decisions, these traits can be categorized into 3 groups: (a) high directional selection, (b) moderate selection, and (c) intermediate optimum selection. The first 2 categories can be explained by positive assortative mating, and the last can be explained by negative assortative or disassortative mating. Genetic progress was defined by genetic gain per year based on average standardized genomic predictions for cows from 2000 to 2014. Traits with more genetic progress tended to have more "inflated" genomic predictions (i.e., "inflation" means here that genomic predictions are larger in absolute values than expected, whereas "deflation" means smaller than expected). Heritability estimates for 14 out of 18 traits declined in the last 16 yr, and Δh2 ranged from -0.09 to 0.04. Traits with a greater decline in heritability tended to have more deflated genomic predictions. Biases (inflation or deflation) in genomic predictions were not improved by using the latest genetic parameters, implying that bias in genomic predictions due to preselection was not substantial for a large-scale genomic evaluation. Moreover, the strong selection intensity was not fully responsible for bias in genomic predictions. The directional selection can decrease heritability; however, positive assortative mating, which was strongly associated with large genetic gains, could minimize the decline in heritability for a trait under strong selection and could affect bias in genomic predictions.

Impact of phenotypic information and composition of reference population on genomic prediction in fish under the presence of genotype by environment interaction

The difficulty or high cost to register pedigree and the presence of genotype by environment interaction (G×E) are the factors limiting conventional selective breeding in aquaculture. The limitation could be overcome by genomic selection. The objective of this study was to investigate the impact of phenotypic information and composition of reference population on genomic prediction in aquaculture under the presence of G×E, and the impact of genotyping strategy for selection candidates on genetic progress. The study was based on simulated data mimicking a rainbow trout population under G×E where genetic correlation between breeding environment and commercial production environment was 0.8, 0.5 or 0.2. Different scenarios of reference population were investigated, with regard to reference individuals from breeding population or production population, reference individuals were a random sample or a selected sample of top fish, and the phenotypic performance of the reference individuals were measured in breeding environment or in production environment. These scenarios were assessed on accuracy and unbiasedness of predicted breeding value for performance in commercial production environment. The results showed that genomic prediction greatly increased accuracy of selection, compared with conventional pedigree-based selection. Reference population comprising top individuals led to low accuracy and serious bias of genomic prediction. Accuracies of breeding values for performance in production environment predicted based on phenotypic information from breeding farm were low when genetic correlation between breeding and production environments was less than 0.8. With records of performance measured in production farm, reference population comprising individuals from breeding population gave higher prediction accuracy than the reference individuals from production population did. In addition, preselecting candidates from the breeding population for genotyping led to larger selection response than randomly genotyping candidates. These results indicate that genomic selection is feasible and promising in fish breeding under the presence of G×E. It is a good strategy to use a reference population comprising individuals randomly sampled from breeding population but reared in commercial production farm to get phenotypic records (RanBC in the context). Such a reference population is representative of the whole population, have a close relationship with candidates, and lead to a direct selection for performance in commercial production environment. In the situation of limited cost for genotyping selection candidates, genotyping top selection candidates is a good strategy since it can realize high selection intensity.

Controlling bias in genomic breeding values for young genotyped bulls

The objectives of this study were to investigate bias in genomic predictions for dairy cattle and to find a practical approach to reduce the bias. The simulated data included phenotypes, pedigrees, and genotypes, mimicking a dairy cattle population (i.e., cows with phenotypes and bulls with no phenotypes) and assuming selection by breeding values or no selection. With the simulated data, genomic estimated breeding values (GEBV) were calculated with a single-step genomic BLUP and compared with true breeding values. Phenotypes and genotypes were simulated in 10 generations and in the last 4 generations, respectively. Phenotypes in the last generation were removed to predict breeding values for those individuals using only genomic and pedigree information. Complete pedigrees and incomplete pedigrees with 50% missing dams were created to construct the pedigree-based relationship matrix with and without inbreeding. With missing dams, unknown parent groups (UPG) were assigned in relationship matrices. Regression coefficients (b1) and coefficients of determination (R2) of true breeding values on (G)EBV were calculated to investigate inflation and accuracy in GEBV for genotyped animals, respectively. In addition to the simulation study, 18 linear type traits of US Holsteins were examined. For the 18 type traits, b1 and R2 of GEBV with full data sets on GEBV with partial data sets for young genotyped bulls were calculated. The results from the simulation study indicated inflation in GEBV for genotyped males that were evaluated with only pedigree and genomic information under BLUP selection. However, when UPG for only pedigree-based relationships were included, the inflation was reduced, accuracy was highest, and genetic trends had no bias. For the linear type traits, when UPG for only pedigree-based relationships were included, the results were generally in agreement with those from the simulation study, implying less bias in genetic trends. However, when including no UPG, UPG in pedigree-based relationships, or UPG in genomic relationships, inflation and accuracy in GEBV were similar. The results from the simulation and type traits suggest that UPG must be defined accurately to be estimable and inbreeding should be included in pedigree-based relationships. In dairy cattle, known pedigree information with inbreeding and estimable UPG plays an important role in improving compatibility between pedigree-based and genomic relationship matrices, resulting in more reliable genomic predictions.

Accounting for population structure in selective cow genotyping strategies.

The objective of the present study was to investigate the impact of considering population structure in cow genotyping strategies over the accuracy and bias of genomic predictions. A small dairy cattle population was simulated to address these objectives. Based on four main traditional designs (random, top-yield, extreme-yield and top-accuracy cows), different numbers (1,000; 2,000 and 5,000) of cows were sampled and included in the reference population. Traditional designs were replicated considering or not population structure and compared among and with a reference population containing only bulls. The inclusion of cows increased accuracy in all scenarios compared with using only bulls. Scenarios accounting for population structure when choosing cows to the reference population slightly outperformed their traditional versions by yielding higher accuracy and lower bias in genomic predictions. Building a cow-based reference population from groups of related individuals considering the frequency of individuals from those same groups in the validation population yielded promising results with applications on selection for expensive- or difficult-to-measure traits. Methods here presented may be easily implemented in both new or already established breeding programs, as they improved prediction and reduced bias in genomic evaluations while demanding no additional costs.

A comprehensive comparison between single- and two-step GBLUP methods in a simulated beef cattle population

The statistical methods used in the genetic evaluations are a key component of the process and can be best compared by using simulated data. The latter is especially true in grazing beef cattle production systems, where the number of proven bulls with highly reliable estimated breeding values is limited to allow for a trustworthy validation of genomic predictions. Therefore, we simulated data for 4980 beef cattle aiming to compare single-step genomic best linear unbiased prediction (ssGBLUP), which simultaneously incorporates pedigree, phenotypic, and genomic data into genomic evaluations, and two-step GBLUP (tsGBLUP) procedures and genomic estimated breeding values (GEBVs) blending methods. The greatest increases in GEBV accuracies compared with the parents’ average estimated breeding values (EBVPA) were 0.364 and 0.341 for ssGBLUP and tsGBLUP, respectively. Direct genomic value and GEBV accuracies when using ssGBLUP and tsGBLUP procedures were similar, except for the GEBV accuracies using Hayes’ blending method in tsGBLUP. There was no significant or slight bias in genomic predictions from ssGBLUP or tsGBLUP (using VanRaden’s blending method), indicating that these predictions are on the same scale compared with the true breeding values. Overall, genetic evaluations including genomic information resulted in gains in accuracy >100% compared with the EBVPA. In addition, there were no significant differences between the selected animals (10% males and 50% females) by using ssGBLUP or tsGBLUP.

Effects of preselection of genotyped animals on reliability and bias of genomic prediction in dairy cattle.

ObjectiveModels for genomic selection assume that the reference population is an unselected population. However, in practice, genotyped individuals, such as progeny-tested bulls, are highly selected, and the reference population is created after preselection. In dairy cattle, the intensity of selection is higher in males than in females, suggesting that cows can be added to the reference population with less bias and loss of accuracy. The objective is to develop formulas applied to any genomic prediction studies or practice with preselected animals as reference population.MethodsWe developed formulas for calculating the reliability and bias of genomically enhanced breeding values (GEBV) in the reference population where individuals are preselected on estimated breeding values. Based on the formulas presented, deterministic simulation was conducted by varying heritability, preselection percentage, and the reference population size.ResultsThe number of bulls equal to a cow regarding the reliability of GEBV was expressed through a simple formula for the reference population consisting of preselected animals. The bull population was vastly superior to the cow population regarding the reliability of GEBV for low-heritability traits. However, the superiority of reliability from the bull reference population over the cow population decreased as heritability increased. Bias was greater for bulls than cows. Bias and reduction in reliability of GEBV due to preselection was alleviated by expanding reference population.ConclusionCows are easier in expanding reference population size compared with bulls and alleviate bias and reduction in reliability of GEBV of bulls which are highly preselected than cows by expanding the cow reference population.

The effect of using cow genomic information on accuracy and bias of genomic breeding values in a simulated Holstein dairy cattle population

Using cow data in the training population is attractive as a way to mitigate bias due to highly selected training bulls and to implement genomic selection for countries with no or limited proven bull data. However, one potential issue with cow data is a bias due to the preferential treatment. The objectives of this study were to (1) investigate the effect of including cow genotype and phenotype data into the training population on accuracy and bias of genomic predictions and (2) assess the effect of preferential treatment for different proportions of elite cows. First, a 4-pathway Holstein dairy cattle population was simulated for 2 traits with low (0.05) and moderate (0.3) heritability. Then different numbers of cows (0, 2,500, 5,000, 10,000, 15,000, or 20,000) were randomly selected and added to the training group composed of different numbers of top bulls (0, 2,500, 5,000, 10,000, or 15,000). Reliability levels of de-regressed estimated breeding values for training cows and bulls were 30 and 75% for traits with low heritability and were 60 and 90% for traits with moderate heritability, respectively. Preferential treatment was simulated by introducing upward bias equal to 35% of phenotypic variance to 5, 10, and 20% of elite bull dams in each scenario. Two different validation data sets were considered: (1) all animals in the last generation of both elite and commercial tiers (n = 42,000) and (2) only animals in the last generation of the elite tier (n = 12,000). Adding cow data into the training population led to an increase in accuracy (r) and decrease in bias of genomic predictions in all considered scenarios without preferential treatment. The gain in r was higher for the low heritable trait (from 0.004 to 0.166 r points) compared with the moderate heritable trait (from 0.004 to 0.116 r points). The gain in accuracy in scenarios with a lower number of training bulls was relatively higher (from 0.093 to 0.166 r points) than with a higher number of training bulls (from 0.004 to 0.09 r points). In this study, as expected, the bull-only reference population resulted in higher accuracy compared with the cow-only reference population of the same size. However, the cow reference population might be an option for countries with a small-scale progeny testing scheme or for minor breeds in large counties, and for traits measured only on a small fraction of the population. The inclusion of preferential treatment to 5 to 20% of the elite cows led to an adverse effect on both accuracy and bias of predictions. When preferential treatment was present, random selection of cows did not reduce the effect of preferential treatment.

Accuracy and bias of genomic prediction with different de-regression methods

Genomic selection has become increasingly important in the breeding of animals and plants. The response variable is an important factor, influencing the accuracy of genomic selection. The de-regressed proof (DRP) based on traditional estimated breeding value (EBV) is commonly used as response variable. In the current study, simulated data from 16th QTL-MAS Workshop and real data from Chinese Holstein cattle were used to compare accuracy and bias of genomic prediction with two methods of calculating DRP. Our results with simulated data showed that the correlation between genomic EBV and true breeding value achieved using the Jairath method (DRP_J) was superior to that achieved using the Garrick method (DRP_G) for simulated trait 1 but the reverse was true for simulated trait 3, and these two methods performed comparably for simulated trait 2. For all three simulated traits, DRP_J yielded larger bias of genomic prediction. However, DRP_J outperformed DRP_G in both accuracy and unbiasedness for four milk production traits in Chinese Holstein. In the estimation of genomic breeding value using genomic BLUP model, two methods for weighting diagonal elements of incidence matrix associated with residual error were also compared. With increasing the proportion of genetic variance unexplained by markers, the accuracy of genomic prediction was decreased and the bias was increased. Weighting by the reliability of DRP produced accuracy comparable to the evaluation where the proportion of genetic variance unexplained by markers was considered, but with smaller bias in general.

Comparison of genomic breeding values of Holstein in Czech Republic

Genomic evaluation by single-step procedure allows efficient implanting of genetic SNP markers into complicated procedure of random regression test-day-model of milk traits. Test-day records and pedigree files on nation-wide scale are combined with genomic relationship into common evaluation of both genotyped and ungenotyped animals. Due to strong import of foreign sperm into small national populations is the reliability of evaluation of young animals low. This is particularly seen in evaluation of young bulls, which frequently have both parents foreign. Genomic evaluation helps and notably improves reliability of evaluation. ssGBLUP procedure is advantageous especially for small populations. Domestic genomic evaluation of young animals has medium to high correlation with foreign Interbull values. Interbull conversion of values of bulls according MACE, which works with progeny tested bulls, is more reliable than conversion according GMACE procedure, which works with genomic evaluation of young animals. Keywords: genomic breeding value, ssGBLUP, test day model, MACE, GMACE References Bauer, J. et al. (2014) Approximation of the reliability of single-step genomic breeding values for dairy cattle in the Czech Republic. Anim. Sci. Papers and Reports, 32, pp. 301-306. Bauer, J., Přibyl, J. and Vostrý, L. (2015) Contribution of domestic production records and Interbull EBV on approximate reliabilities of single-step genomic breeding values in dairy cattle. Czech J. Anim. Sci., 60, 263-267. Christensen, O.F. and Lund, M.S. (2010) Genomic prediction when some animals are not genotyped. Genet.Sel.Evol. 42, pp. 2. Forni, S., Aguilar I. and Misztal, I. (2011) Different genomic relationship matrices for single step analysis using phenotypic, pedigree and genomic information. Genet. Sel. Evol., 43, pp. 1. Legarra A., Aguilar I. and Misztal, I. (2009) A relationship matrix including full pedigree and genomic information. J. Dairy Sci., 92, pp. 4656-4663. Meuwissen, T.H.E., Hayes, B.J. and Goddard, M.E. (2001) Prediction of total genetic value using genome-wide dense marker maps. Genetics, 157, pp. 1819–1829. Misztal, I., Legarra A. and Aguilar, I. (2009) Computing procedures for genetic evaluation including phenotypic, full pedigree, and genomic information. J. Dairy Sci., 92, pp. 4648–4655. Pesek, P., Přibyl, J. and Vostrý, L . (2015) Genetic variances of SNP loci for milk yield in dairy cattle. J. Appl. Genet., 56, pp. 339-347. Plemdat, (2015) Descriptions of Breeding values Evaluation. Retrieved on 10 th June 2015 From www.plemdat.cz. Přibyl, J. et al. (2014) Domestic and Interbull information in the single step genomic evaluation of Holstein milk production. Czech J. Anim. Sci., 59, pp. 409-415. Přibyl, J. et al. (2015) Domestic estimated breeding values and genomic enhanced breeding values of bulls in comparison with their foreign genomic enhanced breeding values. Animal, 9, pp. 1635-1642. Vitezica, Z.G. et al. (2011) Bias in genomic predictions for populations under selection. Genet. Res. (Camb), 93, pp. 357-366. Zavadilova, L., Jamrozik, J. and Schaeffer, L.R. (2005a) Genetic parameters for test- day model with random regressions for production traits of Czech Holstein cattle. Czech J. Anim. Sci., 50, pp. 142-154. Zavadilova, L., Němcova, E. and Wolf, J. (2005b) Definition of subgroups for fixed regression in the test-day animal model for milk production of Holstein cattle in the Czech Republic. Czech J. Anim. Sci., 50, pp. 7-13. Zavadilova, L. et al. (2014) Single-step genomic evaluation for linear type traits of Holstein cows in Czech Republic. Anim. Sci. Papers and Reports vol. 32, pp. 201-208.

Including different groups of genotyped females for genomic prediction in a Nordic Jersey population

Including genotyped females in a reference population (RP) is an obvious way to increase the RP in genomic selection, especially for dairy breeds of limited population size. However, the incorporation of these females must be conducted cautiously because of the potential preferential treatment of the genotyped cows and lower reliabilities of phenotypes compared with the proven pseudo-phenotypes of bulls. Breeding organizations in Denmark, Finland, and Sweden have implemented a female-genotyping project with the possibility of genotyping entire herds using the low-density (LD) chip. In the present study, 5 scenarios for building an RP were investigated in the Nordic Jersey population: (1) bulls only, (2) bulls with females from the LD project, (3) bulls with females from the LD project plus non-LD project females genotyped before their first calving, (4) bulls with females from the LD project plus non-LD project females genotyped after their first calving, and (5) bulls with all genotyped females. The genomically enhanced breeding value (GEBV) was predicted for 8 traits in the Nordic total merit index through a genomic BLUP model using deregressed proof (DRP) as the response variable in all scenarios. In addition, (daughter) yield deviation and raw phenotypic data were studied as response variables for comparison with the DRP, using stature as a model trait. The validation population was formed using a cut-off birth year of 2005 based on the genotyped Nordic Jersey bulls with DRP. The average increment in reliability of the GEBV across the 8 traits investigated was 1.9 to 4.5 percentage points compared with using only bulls in the RP (scenario 1). The addition of all the genotyped females to the RP resulted in the highest gain in reliability (scenario 5), followed by scenario 3, scenario 2, and scenario 4. All scenarios led to inflated GEBV because the regression coefficients are less than 1. However, scenario 2 and scenario 3 led to less bias of genomic predictions than scenario 5, with regression coefficients showing less deviation from scenario 1. For the study on stature, the daughter yield deviation/daughter yield deviation performed slightly better than the DRP as the response variable in the genomic BLUP (GBLUP) model. Therefore, adding unselected females in the RP could significantly improve the reliabilities and tended to reduce the prediction bias compared with adding selectively genotyped females. Although the DRP has performed robustly so far, the use of raw data is recommended with a single-step model as an optimal solution for future genomic evaluations.

Genomic Prediction in Animals and Plants: Simulation of Data, Validation, Reporting, and Benchmarking

The genomic prediction of phenotypes and breeding values in animals and plants has developed rapidly into its own research field. Results of genomic prediction studies are often difficult to compare because data simulation varies, real or simulated data are not fully described, and not all relevant results are reported. In addition, some new methods have been compared only in limited genetic architectures, leading to potentially misleading conclusions. In this article we review simulation procedures, discuss validation and reporting of results, and apply benchmark procedures for a variety of genomic prediction methods in simulated and real example data. Plant and animal breeding programs are being transformed by the use of genomic data, which are becoming widely available and cost-effective to predict genetic merit. A large number of genomic prediction studies have been published using both simulated and real data. The relative novelty of this area of research has made the development of scientific conventions difficult with regard to description of the real data, simulation of genomes, validation and reporting of results, and forward in time methods. In this review article we discuss the generation of simulated genotype and phenotype data, using approaches such as the coalescent and forward in time simulation. We outline ways to validate simulated data and genomic prediction results, including cross-validation. The accuracy and bias of genomic prediction are highlighted as performance indicators that should be reported. We suggest that a measure of relatedness between the reference and validation individuals be reported, as its impact on the accuracy of genomic prediction is substantial. A large number of methods were compared in example simulated and real (pine and wheat) data sets, all of which are publicly available. In our limited simulations, most methods performed similarly in traits with a large number of quantitative trait loci (QTL), whereas in traits with fewer QTL variable selection did have some advantages. In the real data sets examined here all methods had very similar accuracies. We conclude that no single method can serve as a benchmark for genomic prediction. We recommend comparing accuracy and bias of new methods to results from genomic best linear prediction and a variable selection approach (e.g., BayesB), because, together, these methods are appropriate for a range of genetic architectures. An accompanying article in this issue provides a comprehensive review of genomic prediction methods and discusses a selection of topics related to application of genomic prediction in plants and animals.

Comparison on genomic predictions using three GBLUP methods and two single-step blending methods in the Nordic Holstein population

BackgroundA single-step blending approach allows genomic prediction using information of genotyped and non-genotyped animals simultaneously. However, the combined relationship matrix in a single-step method may need to be adjusted because marker-based and pedigree-based relationship matrices may not be on the same scale. The same may apply when a GBLUP model includes both genomic breeding values and residual polygenic effects. The objective of this study was to compare single-step blending methods and GBLUP methods with and without adjustment of the genomic relationship matrix for genomic prediction of 16 traits in the Nordic Holstein population.MethodsThe data consisted of de-regressed proofs (DRP) for 5 214 genotyped and 9 374 non-genotyped bulls. The bulls were divided into a training and a validation population by birth date, October 1, 2001. Five approaches for genomic prediction were used: 1) a simple GBLUP method, 2) a GBLUP method with a polygenic effect, 3) an adjusted GBLUP method with a polygenic effect, 4) a single-step blending method, and 5) an adjusted single-step blending method. In the adjusted GBLUP and single-step methods, the genomic relationship matrix was adjusted for the difference of scale between the genomic and the pedigree relationship matrices. A set of weights on the pedigree relationship matrix (ranging from 0.05 to 0.40) was used to build the combined relationship matrix in the single-step blending method and the GBLUP method with a polygenetic effect.ResultsAveraged over the 16 traits, reliabilities of genomic breeding values predicted using the GBLUP method with a polygenic effect (relative weight of 0.20) were 0.3% higher than reliabilities from the simple GBLUP method (without a polygenic effect). The adjusted single-step blending and original single-step blending methods (relative weight of 0.20) had average reliabilities that were 2.1% and 1.8% higher than the simple GBLUP method, respectively. In addition, the GBLUP method with a polygenic effect led to less bias of genomic predictions than the simple GBLUP method, and both single-step blending methods yielded less bias of predictions than all GBLUP methods.ConclusionsThe single-step blending method is an appealing approach for practical genomic prediction in dairy cattle. Genomic prediction from the single-step blending method can be improved by adjusting the scale of the genomic relationship matrix.

Bias in genomic predictions for populations under selection

Prediction of genetic merit or disease risk using genetic marker information is becoming a common practice for selection of livestock and plant species. For the successful application of genome-wide marker-assisted selection (GWMAS), genomic predictions should be accurate and unbiased. The effect of selection on bias and accuracy of genomic predictions was studied in two simulated animal populations under weak or strong selection and with several heritabilities. Prediction of genetic values was by best-linear unbiased prediction (BLUP) using data either from relatives summarized in pseudodata for genotyped individuals (multiple-step method) or using all available data jointly (single-step method). The single-step method combined genomic- and pedigree-based relationship matrices. Predictions by the multiple-step method were biased. Predictions by a single-step method were less biased and more accurate but under strong selection were less accurate. When genomic relationships were shifted by a constant, the single-step method was unbiased and the most accurate. The value of that constant, which adjusts for non-random selection of genotyped individuals, can be derived analytically.