A comparison of five methods to predict genomic breeding values of dairy bulls from genome-wide SNP markers

Background

Genomic selection (GS) uses molecular breeding values (MBV) derived from dense markers across the entire genome for selection of young animals. The accuracy of MBV prediction is important for a successful application of GS. Recently, several methods have been proposed to estimate MBV. Initial simulation studies have shown that these methods can accurately predict MBV. In this study we compared the accuracies and possible bias of five different regression methods in an empirical application in dairy cattle.

Methods

Genotypes of 7,372 SNP and highly accurate EBV of 1,945 dairy bulls were used to predict MBV for protein percentage (PPT) and a profit index (Australian Selection Index, ASI). Marker effects were estimated by least squares regression (FR-LS), Bayesian regression (Bayes-R), random regression best linear unbiased prediction (RR-BLUP), partial least squares regression (PLSR) and nonparametric support vector regression (SVR) in a training set of 1,239 bulls. Accuracy and bias of MBV prediction were calculated from cross-validation of the training set and tested against a test team of 706 young bulls.

Results

For both traits, FR-LS using a subset of SNP was significantly less accurate than all other methods which used all SNP. Accuracies obtained by Bayes-R, RR-BLUP, PLSR and SVR were very similar for ASI (0.39-0.45) and for PPT (0.55-0.61). Overall, SVR gave the highest accuracy.All methods resulted in biased MBV predictions for ASI, for PPT only RR-BLUP and SVR predictions were unbiased. A significant decrease in accuracy of prediction of ASI was seen in young test cohorts of bulls compared to the accuracy derived from cross-validation of the training set. This reduction was not apparent for PPT. Combining MBV predictions with pedigree based predictions gave 1.05 - 1.34 times higher accuracies compared to predictions based on pedigree alone. Some methods have largely different computational requirements, with PLSR and RR-BLUP requiring the least computing time.

Conclusions

The four methods which use information from all SNP namely RR-BLUP, Bayes-R, PLSR and SVR generate similar accuracies of MBV prediction for genomic selection, and their use in the selection of immediate future generations in dairy cattle will be comparable. The use of FR-LS in genomic selection is not recommended.     

The accuracy of prediction of genomic selection in elite hybrid rye populations surpasses the accuracy of marker-assisted selection and is equally augmented by multiple field evaluation locations and test years

Background

Marker-assisted selection (MAS) and genomic selection (GS) based on genome-wide marker data provide powerful tools to predict the genotypic value of selection material in plant breeding. However, case-to-case optimization of these approaches is required to achieve maximum accuracy of prediction with reasonable input.

Results

Based on extended field evaluation data for grain yield, plant height, starch content and total pentosan content of elite hybrid rye derived from testcrosses involving two bi-parental populations that were genotyped with 1048 molecular markers, we compared the accuracy of prediction of MAS and GS in a cross-validation approach. MAS delivered generally lower and in addition potentially over-estimated accuracies of prediction than GS by ridge regression best linear unbiased prediction (RR-BLUP). The grade of relatedness of the plant material included in the estimation and test sets clearly affected the accuracy of prediction of GS. Within each of the two bi-parental populations, accuracies differed depending on the relatedness of the respective parental lines. Across populations, accuracy increased when both populations contributed to estimation and test set. In contrast, accuracy of prediction based on an estimation set from one population to a test set from the other population was low despite that the two bi-parental segregating populations under scrutiny shared one parental line. Limiting the number of locations or years in field testing reduced the accuracy of prediction of GS equally, supporting the view that to establish robust GS calibration models a sufficient number of test locations is of similar importance as extended testing for more than one year.

Conclusions

In hybrid rye, genomic selection is superior to marker-assisted selection. However, it achieves high accuracies of prediction only for selection candidates closely related to the plant material evaluated in field trials, resulting in a rather pessimistic prognosis for distantly related material. Both, the numbers of evaluation locations and testing years in trials contribute equally to prediction accuracy.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-556) contains supplementary material, which is available to authorized users.     

Empirical and deterministic accuracies of across-population genomic prediction

Background

Differences in linkage disequilibrium and in allele substitution effects of QTL (quantitative trait loci) may hinder genomic prediction across populations. Our objective was to develop a deterministic formula to estimate the accuracy of across-population genomic prediction, for which reference individuals and selection candidates are from different populations, and to investigate the impact of differences in allele substitution effects across populations and of the number of QTL underlying a trait on the accuracy.

Methods

A deterministic formula to estimate the accuracy of across-population genomic prediction was derived based on selection index theory. Moreover, accuracies were deterministically predicted using a formula based on population parameters and empirically calculated using simulated phenotypes and a GBLUP (genomic best linear unbiased prediction) model. Phenotypes of 1033 Holstein-Friesian, 105 Groninger White Headed and 147 Meuse-Rhine-Yssel cows were simulated by sampling 3000, 300, 30 or 3 QTL from the available high-density SNP (single nucleotide polymorphism) information of three chromosomes, assuming a correlation of 1.0, 0.8, 0.6, 0.4, or 0.2 between allele substitution effects across breeds. The simulated heritability was set to 0.95 to resemble the heritability of deregressed proofs of bulls.

Results

Accuracies estimated with the deterministic formula based on selection index theory were similar to empirical accuracies for all scenarios, while accuracies predicted with the formula based on population parameters overestimated empirical accuracies by ~25 to 30%. When the between-breed genetic correlation differed from 1, i.e. allele substitution effects differed across breeds, empirical and deterministic accuracies decreased in proportion to the genetic correlation. Using a multi-trait model, it was possible to accurately estimate the genetic correlation between the breeds based on phenotypes and high-density genotypes. The number of QTL underlying the simulated trait did not affect the accuracy.

Conclusions

The deterministic formula based on selection index theory estimated the accuracy of across-population genomic predictions well. The deterministic formula using population parameters overestimated the across-population genomic accuracy, but may still be useful because of its simplicity. Both formulas could accommodate for genetic correlations between populations lower than 1. The number of QTL underlying a trait did not affect the accuracy of across-population genomic prediction using a GBLUP method.     

Response and inbreeding from a genomic selection experiment in layer chickens

Background

Genomic selection (GS) using estimated breeding values (GS-EBV) based on dense marker data is a promising approach for genetic improvement. A simulation study was undertaken to illustrate the opportunities offered by GS for designing breeding programs. It consisted of a selection program for a sex-limited trait in layer chickens, which was developed by deterministic predictions under different scenarios. Later, one of the possible schemes was implemented in a real population of layer chicken.

Methods

In the simulation, the aim was to double the response to selection per year by reducing the generation interval by 50 %, while maintaining the same rate of inbreeding per year. We found that GS with retraining could achieve the set objectives while requiring 75 % fewer reared birds and 82 % fewer phenotyped birds per year. A multi-trait GS scenario was subsequently implemented in a real population of brown egg laying hens. The population was split into two sub-lines, one was submitted to conventional phenotypic selection, and one was selected based on genomic prediction. At the end of the 3-year experiment, the two sub-lines were compared for multiple performance traits that are relevant for commercial egg production.

Results

Birds that were selected based on genomic prediction outperformed those that were submitted to conventional selection for most of the 16 traits that were included in the index used for selection. However, although the two programs were designed to achieve the same rate of inbreeding per year, the realized inbreeding per year assessed from pedigree was higher in the genomic selected line than in the conventionally selected line.

Conclusions

The results demonstrate that GS is a promising alternative to conventional breeding for genetic improvement of layer chickens.     

Genomic selection accuracies within and between environments and small breeding groups in white spruce

Background

Genomic selection (GS) may improve selection response over conventional pedigree-based selection if markers capture more detailed information than pedigrees in recently domesticated tree species and/or make it more cost effective. Genomic prediction accuracies using 1748 trees and 6932 SNPs representative of as many distinct gene loci were determined for growth and wood traits in white spruce, within and between environments and breeding groups (BG), each with an effective size of N_e ≈ 20. Marker subsets were also tested.

Results

Model fits and/or cross-validation (CV) prediction accuracies for ridge regression (RR) and the least absolute shrinkage and selection operator models approached those of pedigree-based models. With strong relatedness between CV sets, prediction accuracies for RR within environment and BG were high for wood (r = 0.71–0.79) and moderately high for growth (r = 0.52–0.69) traits, in line with trends in heritabilities. For both classes of traits, these accuracies achieved between 83% and 92% of those obtained with phenotypes and pedigree information. Prediction into untested environments remained moderately high for wood (r ≥ 0.61) but dropped significantly for growth (r ≥ 0.24) traits, emphasizing the need to phenotype in all test environments and model genotype-by-environment interactions for growth traits. Removing relatedness between CV sets sharply decreased prediction accuracies for all traits and subpopulations, falling near zero between BGs with no known shared ancestry. For marker subsets, similar patterns were observed but with lower prediction accuracies.

Conclusions

Given the need for high relatedness between CV sets to obtain good prediction accuracies, we recommend to build GS models for prediction within the same breeding population only. Breeding groups could be merged to build genomic prediction models as long as the total effective population size does not exceed 50 individuals in order to obtain high prediction accuracy such as that obtained in the present study. A number of markers limited to a few hundred would not negatively impact prediction accuracies, but these could decrease more rapidly over generations. The most promising short-term approach for genomic selection would likely be the selection of superior individuals within large full-sib families vegetatively propagated to implement multiclonal forestry.     

Efficiency of multi-trait,indirect, and trait-assisted genomic selection for improvement of biomass sorghum

Key message

We compare genomic selection methods that use correlated traits to help predict biomass yield in sorghum, and find that trait-assisted genomic selection performs best.

Abstract

Genomic selection (GS) is usually performed on a single trait, but correlated traits can also help predict a focal trait through indirect or multi-trait GS. In this study, we use a pre-breeding population of biomass sorghum to compare strategies that use correlated traits to improve prediction of biomass yield, the focal trait. Correlated traits include moisture, plant height measured at monthly intervals between planting and harvest, and the area under the growth progress curve. In addition to single- and multi-trait direct and indirect GS, we test a new strategy called trait-assisted GS, in which correlated traits are used along with marker data in the validation population to predict a focal trait. Single-trait GS for biomass yield had a prediction accuracy of 0.40. Indirect GS performed best using area under the growth progress curve to predict biomass yield, with a prediction accuracy of 0.37, and did not differ from indirect multi-trait GS that also used moisture information. Multi-trait GS and single-trait GS yielded similar results, indicating that correlated traits did not improve prediction of biomass yield in a standard GS scenario. However, trait-assisted GS increased prediction accuracy by up to \(50\%\) when using plant height in both the training and validation populations to help predict yield in the validation population. Coincidence between selected genotypes in phenotypic and genomic selection was also highest in trait-assisted GS. Overall, these results suggest that trait-assisted GS can be an efficient strategy when correlated traits are obtained earlier or more inexpensively than a focal trait.

     

solGS: a web-based tool for genomic selection

Background

Genomic selection (GS) promises to improve accuracy in estimating breeding values and genetic gain for quantitative traits compared to traditional breeding methods. Its reliance on high-throughput genome-wide markers and statistical complexity, however, is a serious challenge in data management, analysis, and sharing. A bioinformatics infrastructure for data storage and access, and user-friendly web-based tool for analysis and sharing output is needed to make GS more practical for breeders.

Results

We have developed a web-based tool, called solGS, for predicting genomic estimated breeding values (GEBVs) of individuals, using a Ridge-Regression Best Linear Unbiased Predictor (RR-BLUP) model. It has an intuitive web-interface for selecting a training population for modeling and estimating genomic estimated breeding values of selection candidates. It estimates phenotypic correlation and heritability of traits and selection indices of individuals. Raw data is stored in a generic database schema, Chado Natural Diversity, co-developed by multiple database groups. Analysis output is graphically visualized and can be interactively explored online or downloaded in text format. An instance of its implementation can be accessed at the NEXTGEN Cassava breeding database, http://cassavabase.org/solgs.

Conclusions

solGS enables breeders to store raw data and estimate GEBVs of individuals online, in an intuitive and interactive workflow. It can be adapted to any breeding program.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-014-0398-7) contains supplementary material, which is available to authorized users.     

Genomic selection for grain yield and quality traits in durum wheat

The prediction accuracies of genomic selection depend on several factors, including the genetic architecture of target traits, the number of traits considered at a given time, and the statistical models. Here, we assessed the potential of single-trait (ST) and multi-trait (MT) genomic prediction models for durum wheat on yield and quality traits using a breeding panel (BP) of 170 varieties and advanced breeding lines, and a doubled-haploid (DH) population of 154 lines. The two populations were genotyped with the Infinium iSelect 90K SNP assay and phenotyped for various traits. Six ST-GS models (RR-BLUP, G-BLUP, BayesA, BayesB, Bayesian LASSO, and RKHS) and three MT prediction approaches (MT-BayesA, MT-Matrix, and MT-SI approaches which use economic selection index as a trait value) were applied for predicting yield, protein content, gluten index, and alveograph measures. The ST prediction accuracies ranged from 0.5 to 0.8 for the various traits and models and revealed comparable prediction accuracies for most of the traits in both populations, except BayesA and BayesB, which better predicted gluten index, tenacity, and strength in the DH population. The MT-GS models were more accurate than the ST-GS models only for grain yield in the BP. Using BP as a training set to predict the DH population resulted in poor predictions. Overall, all the six ST-GS models appear to be applicable for GS of yield and gluten strength traits in durum wheat, but we recommend the simple computational models RR-BLUP or G-BLUP for predicating single trait and MT-SI for predicting yield and protein simultaneously.     

Genomic prediction based on data from three layer lines: a comparison between linear methods

Background

The prediction accuracy of several linear genomic prediction models, which have previously been used for within-line genomic prediction, was evaluated for multi-line genomic prediction.

Methods

Compared to a conventional BLUP (best linear unbiased prediction) model using pedigree data, we evaluated the following genomic prediction models: genome-enabled BLUP (GBLUP), ridge regression BLUP (RRBLUP), principal component analysis followed by ridge regression (RRPCA), BayesC and Bayesian stochastic search variable selection. Prediction accuracy was measured as the correlation between predicted breeding values and observed phenotypes divided by the square root of the heritability. The data used concerned laying hens with phenotypes for number of eggs in the first production period and known genotypes. The hens were from two closely-related brown layer lines (B1 and B2), and a third distantly-related white layer line (W1). Lines had 1004 to 1023 training animals and 238 to 240 validation animals. Training datasets consisted of animals of either single lines, or a combination of two or all three lines, and had 30 508 to 45 974 segregating single nucleotide polymorphisms.

Results

Genomic prediction models yielded 0.13 to 0.16 higher accuracies than pedigree-based BLUP. When excluding the line itself from the training dataset, genomic predictions were generally inaccurate. Use of multiple lines marginally improved prediction accuracy for B2 but did not affect or slightly decreased prediction accuracy for B1 and W1. Differences between models were generally small except for RRPCA which gave considerably higher accuracies for B2. Correlations between genomic predictions from different methods were higher than 0.96 for W1 and higher than 0.88 for B1 and B2. The greater differences between methods for B1 and B2 were probably due to the lower accuracy of predictions for B1 (~0.45) and B2 (~0.40) compared to W1 (~0.76).

Conclusions

Multi-line genomic prediction did not affect or slightly improved prediction accuracy for closely-related lines. For distantly-related lines, multi-line genomic prediction yielded similar or slightly lower accuracies than single-line genomic prediction. Bayesian variable selection and GBLUP generally gave similar accuracies. Overall, RRPCA yielded the greatest accuracies for two lines, suggesting that using PCA helps to alleviate the “n ≪ p” problem in genomic prediction.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0057-5) contains supplementary material, which is available to authorized users.     

Accuracies of genomically estimated breeding values from pure-breed and across-breed predictions in Australian beef cattle

Background

The major obstacles for the implementation of genomic selection in Australian beef cattle are the variety of breeds and in general, small numbers of genotyped and phenotyped individuals per breed. The Australian Beef Cooperative Research Center (Beef CRC) investigated these issues by deriving genomic prediction equations (PE) from a training set of animals that covers a range of breeds and crosses including Angus, Murray Grey, Shorthorn, Hereford, Brahman, Belmont Red, Santa Gertrudis and Tropical Composite. This paper presents accuracies of genomically estimated breeding values (GEBV) that were calculated from these PE in the commercial pure-breed beef cattle seed stock sector.

Methods

PE derived by the Beef CRC from multi-breed and pure-breed training populations were applied to genotyped Angus, Limousin and Brahman sires and young animals, but with no pure-breed Limousin in the training population. The accuracy of the resulting GEBV was assessed by their genetic correlation to their phenotypic target trait in a bi-variate REML approach that models GEBV as trait observations.

Results

Accuracies of most GEBV for Angus and Brahman were between 0.1 and 0.4, with accuracies for abattoir carcass traits generally greater than for live animal body composition traits and reproduction traits. Estimated accuracies greater than 0.5 were only observed for Brahman abattoir carcass traits and for Angus carcass rib fat. Averaged across traits within breeds, accuracies of GEBV were highest when PE from the pooled across-breed training population were used. However, for the Angus and Brahman breeds the difference in accuracy from using pure-breed PE was small. For the Limousin breed no reasonable results could be achieved for any trait.

Conclusion

Although accuracies were generally low compared to published accuracies estimated within breeds, they are in line with those derived in other multi-breed populations. Thus PE developed by the Beef CRC can contribute to the implementation of genomic selection in Australian beef cattle breeding.     

Dynamics of long-term genomic selection

Background

Simulation and empirical studies of genomic selection (GS) show accuracies sufficient to generate rapid gains in early selection cycles. Beyond those cycles, allele frequency changes, recombination, and inbreeding make analytical prediction of gain impossible. The impacts of GS on long-term gain should be studied prior to its implementation.

Methods

A simulation case-study of this issue was done for barley, an inbred crop. On the basis of marker data on 192 breeding lines from an elite six-row spring barley program, stochastic simulation was used to explore the effects of large or small initial training populations with heritabilities of 0.2 or 0.5, applying GS before or after phenotyping, and applying additional weight on low-frequency favorable marker alleles. Genomic predictions were from ridge regression or a Bayesian analysis.

Results

Assuming that applying GS prior to phenotyping shortened breeding cycle time by 50%, this practice strongly increased early selection gains but also caused the loss of many favorable QTL alleles, leading to loss of genetic variance, loss of GS accuracy, and a low selection plateau. Placing additional weight on low-frequency favorable marker alleles, however, allowed GS to increase their frequency earlier on, causing an initial increase in genetic variance. This dynamic led to higher long-term gain while mitigating losses in short-term gain. Weighted GS also increased the maintenance of marker polymorphism, ensuring that QTL-marker linkage disequilibrium was higher than in unweighted GS.

Conclusions

Losing favorable alleles that are in weak linkage disequilibrium with markers is perhaps inevitable when using GS. Placing additional weight on low-frequency favorable alleles, however, may reduce the rate of loss of such alleles to below that of phenotypic selection. Applying such weights at the beginning of GS implementation is important.     

Accuracy of genomic breeding values in multi-breed dairy cattle populations

Background

Two key findings from genomic selection experiments are 1) the reference population used must be very large to subsequently predict accurate genomic estimated breeding values (GEBV), and 2) prediction equations derived in one breed do not predict accurate GEBV when applied to other breeds. Both findings are a problem for breeds where the number of individuals in the reference population is limited. A multi-breed reference population is a potential solution, and here we investigate the accuracies of GEBV in Holstein dairy cattle and Jersey dairy cattle when the reference population is single breed or multi-breed. The accuracies were obtained both as a function of elements of the inverse coefficient matrix and from the realised accuracies of GEBV.

Methods

Best linear unbiased prediction with a multi-breed genomic relationship matrix (GBLUP) and two Bayesian methods (BAYESA and BAYES_SSVS) which estimate individual SNP effects were used to predict GEBV for 400 and 77 young Holstein and Jersey bulls respectively, from a reference population of 781 and 287 Holstein and Jersey bulls, respectively. Genotypes of 39,048 SNP markers were used. Phenotypes in the reference population were de-regressed breeding values for production traits. For the GBLUP method, expected accuracies calculated from the diagonal of the inverse of coefficient matrix were compared to realised accuracies.

Results

When GBLUP was used, expected accuracies from a function of elements of the inverse coefficient matrix agreed reasonably well with realised accuracies calculated from the correlation between GEBV and EBV in single breed populations, but not in multi-breed populations. When the Bayesian methods were used, realised accuracies of GEBV were up to 13% higher when the multi-breed reference population was used than when a pure breed reference was used. However no consistent increase in accuracy across traits was obtained.

Conclusion

Predicting genomic breeding values using a genomic relationship matrix is an attractive approach to implement genomic selection as expected accuracies of GEBV can be readily derived. However in multi-breed populations, Bayesian approaches give higher accuracies for some traits. Finally, multi-breed reference populations will be a valuable resource to fine map QTL.     

The use of communal rearing of families and DNA pooling in aquaculture genomic selection schemes

Background

Traditional family-based aquaculture breeding programs, in which families are kept separately until individual tagging and most traits are measured on the sibs of the candidates, are costly and require a high level of reproductive control. The most widely used alternative is a selection scheme, where families are reared communally and the candidates are selected based on their own individual measurements of the traits under selection. However, in the latter selection schemes, inclusion of new traits depends on the availability of non-invasive techniques to measure the traits on selection candidates. This is a severe limitation of these schemes, especially for disease resistance and fillet quality traits.

Methods

Here, we present a new selection scheme, which was validated using computer simulations comprising 100 families, among which 1, 10 or 100 were reared communally in groups. Pooling of the DNA from 2000, 20000 or 50000 test individuals with the highest and lowest phenotypes was used to estimate 500, 5000 or 10000 marker effects. One thousand or 2000 out of 20000 candidates were preselected for a growth-like trait. These pre-selected candidates were genotyped, and they were selected on their genome-wide breeding values for a trait that could not be measured on the candidates.

Results

A high accuracy of selection, i.e. 0.60-0.88 was obtained with 20000-50000 test individuals but it was reduced when only 2000 test individuals were used. This shows the importance of having large numbers of phenotypic records to accurately estimate marker effects. The accuracy of selection decreased with increasing numbers of families per group.

Conclusions

This new selection scheme combines communal rearing of families, pre-selection of candidates, DNA pooling and genomic selection and makes multi-trait selection possible in aquaculture selection schemes without keeping families separately until individual tagging is possible. The new scheme can also be used for other farmed species, for which the cost of genotyping test individuals may be high, e.g. if trait heritability is low.     

Accuracies of genomic breeding values in American Angus beef cattle using K-means clustering for cross-validation

Background

Genomic selection is a recently developed technology that is beginning to revolutionize animal breeding. The objective of this study was to estimate marker effects to derive prediction equations for direct genomic values for 16 routinely recorded traits of American Angus beef cattle and quantify corresponding accuracies of prediction.

Methods

Deregressed estimated breeding values were used as observations in a weighted analysis to derive direct genomic values for 3570 sires genotyped using the Illumina BovineSNP50 BeadChip. These bulls were clustered into five groups using K-means clustering on pedigree estimates of additive genetic relationships between animals, with the aim of increasing within-group and decreasing between-group relationships. All five combinations of four groups were used for model training, with cross-validation performed in the group not used in training. Bivariate animal models were used for each trait to estimate the genetic correlation between deregressed estimated breeding values and direct genomic values.

Results

Accuracies of direct genomic values ranged from 0.22 to 0.69 for the studied traits, with an average of 0.44. Predictions were more accurate when animals within the validation group were more closely related to animals in the training set. When training and validation sets were formed by random allocation, the accuracies of direct genomic values ranged from 0.38 to 0.85, with an average of 0.65, reflecting the greater relationship between animals in training and validation. The accuracies of direct genomic values obtained from training on older animals and validating in younger animals were intermediate to the accuracies obtained from K-means clustering and random clustering for most traits. The genetic correlation between deregressed estimated breeding values and direct genomic values ranged from 0.15 to 0.80 for the traits studied.

Conclusions

These results suggest that genomic estimates of genetic merit can be produced in beef cattle at a young age but the recurrent inclusion of genotyped sires in retraining analyses will be necessary to routinely produce for the industry the direct genomic values with the highest accuracy.     

Genomic prediction of breeding values using previously estimated SNP variances

Background

Genomic prediction requires estimation of variances of effects of single nucleotide polymorphisms (SNPs), which is computationally demanding, and uses these variances for prediction. We have developed models with separate estimation of SNP variances, which can be applied infrequently, and genomic prediction, which can be applied routinely.

Methods

SNP variances were estimated with Bayes Stochastic Search Variable Selection (BSSVS) and BayesC. Genome-enhanced breeding values (GEBV) were estimated with RR-BLUP (ridge regression best linear unbiased prediction), using either variances obtained from BSSVS (BLUP-SSVS) or BayesC (BLUP-C), or assuming equal variances for each SNP. Datasets used to estimate SNP variances comprised (1) all animals, (2) 50% random animals (RAN50), (3) 50% best animals (TOP50), or (4) 50% worst animals (BOT50). Traits analysed were protein yield, udder depth, somatic cell score, interval between first and last insemination, direct longevity, and longevity including information from predictors.

Results

BLUP-SSVS and BLUP-C yielded similar GEBV as the equivalent Bayesian models that simultaneously estimated SNP variances. Reliabilities of these GEBV were consistently higher than from RR-BLUP, although only significantly for direct longevity. Across scenarios that used data subsets to estimate GEBV, observed reliabilities were generally higher for TOP50 than for RAN50, and much higher than for BOT50. Reliabilities of TOP50 were higher because the training data contained more ancestors of selection candidates. Using estimated SNP variances based on random or non-random subsets of the data, while using all data to estimate GEBV, did not affect reliabilities of the BLUP models. A convergence criterion of 10⁻⁸ instead of 10⁻¹⁰ for BLUP models yielded similar GEBV, while the required number of iterations decreased by 71 to 90%. Including a separate polygenic effect consistently improved reliabilities of the GEBV, but also substantially increased the required number of iterations to reach convergence with RR-BLUP. SNP variances converged faster for BayesC than for BSSVS.

Conclusions

Combining Bayesian variable selection models to re-estimate SNP variances and BLUP models that use those SNP variances, yields GEBV that are similar to those from full Bayesian models. Moreover, these combined models yield predictions with higher reliability and less bias than the commonly used RR-BLUP model.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0052-x) contains supplementary material, which is available to authorized users.     

Genomic prediction based on data from three layer lines using non-linear regression models

Background

Most studies on genomic prediction with reference populations that include multiple lines or breeds have used linear models. Data heterogeneity due to using multiple populations may conflict with model assumptions used in linear regression methods.

Methods

In an attempt to alleviate potential discrepancies between assumptions of linear models and multi-population data, two types of alternative models were used: (1) a multi-trait genomic best linear unbiased prediction (GBLUP) model that modelled trait by line combinations as separate but correlated traits and (2) non-linear models based on kernel learning. These models were compared to conventional linear models for genomic prediction for two lines of brown layer hens (B1 and B2) and one line of white hens (W1). The three lines each had 1004 to 1023 training and 238 to 240 validation animals. Prediction accuracy was evaluated by estimating the correlation between observed phenotypes and predicted breeding values.

Results

When the training dataset included only data from the evaluated line, non-linear models yielded at best a similar accuracy as linear models. In some cases, when adding a distantly related line, the linear models showed a slight decrease in performance, while non-linear models generally showed no change in accuracy. When only information from a closely related line was used for training, linear models and non-linear radial basis function (RBF) kernel models performed similarly. The multi-trait GBLUP model took advantage of the estimated genetic correlations between the lines. Combining linear and non-linear models improved the accuracy of multi-line genomic prediction.

Conclusions

Linear models and non-linear RBF models performed very similarly for genomic prediction, despite the expectation that non-linear models could deal better with the heterogeneous multi-population data. This heterogeneity of the data can be overcome by modelling trait by line combinations as separate but correlated traits, which avoids the occasional occurrence of large negative accuracies when the evaluated line was not included in the training dataset. Furthermore, when using a multi-line training dataset, non-linear models provided information on the genotype data that was complementary to the linear models, which indicates that the underlying data distributions of the three studied lines were indeed heterogeneous.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0075-3) contains supplementary material, which is available to authorized users.     

Will genomic selection be a practical method for plant breeding?

Background

Genomic selection or genome-wide selection (GS) has been highlighted as a new approach for marker-assisted selection (MAS) in recent years. GS is a form of MAS that selects favourable individuals based on genomic estimated breeding values. Previous studies have suggested the utility of GS, especially for capturing small-effect quantitative trait loci, but GS has not become a popular methodology in the field of plant breeding, possibly because there is insufficient information available on GS for practical use.

Scope

In this review, GS is discussed from a practical breeding viewpoint. Statistical approaches employed in GS are briefly described, before the recent progress in GS studies is surveyed. GS practices in plant breeding are then reviewed before future prospects are discussed.

Conclusions

Statistical concepts used in GS are discussed with genetic models and variance decomposition, heritability, breeding value and linear model. Recent progress in GS studies is reviewed with a focus on empirical studies. For the practice of GS in plant breeding, several specific points are discussed including linkage disequilibrium, feature of populations and genotyped markers and breeding scheme. Currently, GS is not perfect, but it is a potent, attractive and valuable approach for plant breeding. This method will be integrated into many practical breeding programmes in the near future with further advances and the maturing of its theory.Key words: Genomic selection, plant breeding, marker assisted selection, genetic model, linkage disequilibrium     

Improving the accuracy of genomic prediction in Chinese Holstein cattle by using one-step blending

Background

The one-step blending approach has been suggested for genomic prediction in dairy cattle. The core of this approach is to incorporate pedigree and phenotypic information of non-genotyped animals. The objective of this study was to investigate the improvement of the accuracy of genomic prediction using the one-step blending method in Chinese Holstein cattle.

Findings

Three methods, GBLUP (genomic best linear unbiased prediction), original one-step blending with a genomic relationship matrix, and adjusted one-step blending with an adjusted genomic relationship matrix, were compared with respect to the accuracy of genomic prediction for five milk production traits in Chinese Holstein. For the two one-step blending methods, de-regressed proofs of 17 509 non-genotyped cows, including 424 dams and 17 085 half-sisters of the validation cows, were incorporated in the prediction model. The results showed that, averaged over the five milk production traits, the one-step blending increased the accuracy of genomic prediction by about 0.12 compared to GBLUP. No further improvement in accuracies was obtained from the adjusted one-step blending over the original one-step blending in our situation. Improvements in accuracies obtained with both one-step blending methods were almost completely contributed by the non-genotyped dams.

Conclusions

Compared with GBLUP, the one-step blending approach can significantly improve the accuracy of genomic prediction for milk production traits in Chinese Holstein cattle. Thus, the one-step blending is a promising approach for practical genomic selection in Chinese Holstein cattle, where the reference population mainly consists of cows.     

Evaluation of genome-wide selection efficiency in maize nested association mapping populations

In comparison to conventional marker-assisted selection (MAS), which utilizes only a subset of genetic markers associated with a trait to predict breeding values (BVs), genome-wide selection (GWS) improves prediction accuracies by incorporating all markers into a model simultaneously. This strategy avoids risks of missing quantitative trait loci (QTL) with small effects. Here, we evaluated the accuracy of prediction for three corn flowering traits days to silking, days to anthesis, and anthesis-silking interval with GWS based on cross-validation experiments using a large data set of 25 nested association mapping populations in maize (Zea mays). We found that GWS via ridge regression-best linear unbiased prediction (RR-BLUP) gave significantly higher predictions compared to MAS utilizing composite interval mapping (CIM). The CIM method may be selected over multiple linear regression to decrease over-estimations of the efficiency of GWS over a MAS strategy. The RR-BLUP method was the preferred method for estimating marker effects in GWS with prediction accuracies comparable to or greater than BayesA and BayesB. The accuracy with RR-BLUP increased with training sample proportion, marker density, and heritability until it reached a plateau. In general, gains in accuracy with RR-BLUP over CIM increased with decreases of these factors. Compared to training sample proportion, the accuracy of prediction with RR-BLUP was relatively insensitive to marker density.     

Accuracy of genomic predictions in Bos indicus (Nellore) cattle

Background

Nellore cattle play an important role in beef production in tropical systems and there is great interest in determining if genomic selection can contribute to accelerate genetic improvement of production and fertility in this breed. We present the first results of the implementation of genomic prediction in a Bos indicus (Nellore) population.

Methods

Influential bulls were genotyped with the Illumina Bovine HD chip in order to assess genomic predictive ability for weight and carcass traits, gestation length, scrotal circumference and two selection indices. 685 samples and 320 238 single nucleotide polymorphisms (SNPs) were used in the analyses. A forward-prediction scheme was adopted to predict the genomic breeding values (DGV). In the training step, the estimated breeding values (EBV) of bulls were deregressed (dEBV) and used as pseudo-phenotypes to estimate marker effects using four methods: genomic BLUP with or without a residual polygenic effect (GBLUP20 and GBLUP0, respectively), a mixture model (Bayes C) and Bayesian LASSO (BLASSO). Empirical accuracies of the resulting genomic predictions were assessed based on the correlation between DGV and dEBV for the testing group.

Results

Accuracies of genomic predictions ranged from 0.17 (navel at weaning) to 0.74 (finishing precocity). Across traits, Bayesian regression models (Bayes C and BLASSO) were more accurate than GBLUP. The average empirical accuracies were 0.39 (GBLUP0), 0.40 (GBLUP20) and 0.44 (Bayes C and BLASSO). Bayes C and BLASSO tended to produce deflated predictions (i.e. slope of the regression of dEBV on DGV greater than 1). Further analyses suggested that higher-than-expected accuracies were observed for traits for which EBV means differed significantly between two breeding subgroups that were identified in a principal component analysis based on genomic relationships.

Conclusions

Bayesian regression models are of interest for future applications of genomic selection in this population, but further improvements are needed to reduce deflation of their predictions. Recurrent updates of the training population would be required to enable accurate prediction of the genetic merit of young animals. The technical feasibility of applying genomic prediction in a Bos indicus (Nellore) population was demonstrated. Further research is needed to permit cost-effective selection decisions using genomic information.