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.     

Association mapping in an elite maize breeding population

Association mapping (AM) is a powerful approach to dissect the genetic architecture of quantitative traits. The main goal of our study was to empirically compare several statistical methods of AM using data of an elite maize breeding program with respect to QTL detection power and possibility to correct for population stratification. These models were based on the inclusion of cofactors (Model A), cofactors and population effect (Model B), and SNP effects nested within populations (Model C). A total of 930 testcross progenies of an elite maize breeding population were field-evaluated for grain yield and grain moisture in multi-location trials and fingerprinted with 425 SNP markers. For grain yield, population stratification was effectively controlled by Model A. For grain moisture with a high ratio of variance among versus within populations, Model B should be applied in order to avoid potential false positives. Model C revealed large differences among allele substitution effects for trait-associated SNPs across multiple plant breeding populations. This heterogeneous SNP allele substitution effects have a severe impact for genomic selection studies, where SNP effects are often assumed to be independent of the genetic background.     

Accuracy of genomic selection to predict maize single-crosses obtained through different mating designs

Key message

Testcross is the worst mating design to use as a training set to predict maize single-crosses that would be obtained through full diallel or North Carolina design II.

Abstract

Even though many papers have been published about genomic prediction (GP) in maize, the best mating design to build the training population has not been defined yet. Such design must maximize the accuracy given constraints on costs and on the logistics of the crosses to be made. Hence, the aims of this work were: (1) empirically evaluate the effect of the mating designs, used as training set, on genomic selection to predict maize single-crosses obtained through full diallel and North Carolina design II, (2) and identify the possibility of reducing the number of crosses and parents to compose these training sets. Our results suggest that testcross is the worst mating design to use as a training set to predict maize single-crosses that would be obtained through full diallel or North Carolina design II. Moreover, North Carolina design II is the best training set to predict hybrids taken from full diallel. However, hybrids from full diallel and North Carolina design II can be well predicted using optimized training sets, which also allow reducing the total number of crosses to be made. Nevertheless, the number of parents and the crosses per parent in the training sets should be maximized.

     

Genome-wide association and genomic selection in animal breeding

Results from genome-wide association studies in livestock, and humans, has lead to the conclusion that the effect of individual quantitative trait loci (QTL) on complex traits, such as yield, are likely to be small; therefore, a large number of QTL are necessary to explain genetic variation in these traits. Given this genetic architecture, gains from marker-assisted selection (MAS) programs using only a small number of DNA markers to trace a limited number of QTL is likely to be small. This has lead to the development of alternative technology for using the available dense single nucleotide polymorphism (SNP) information, called genomic selection. Genomic selection uses a genome-wide panel of dense markers so that all QTL are likely to be in linkage disequilibrium with at least one SNP. The genomic breeding values are predicted to be the sum of the effect of these SNPs across the entire genome. In dairy cattle breeding, the accuracy of genomic estimated breeding values (GEBV) that can be achieved and the fact that these are available early in life have lead to rapid adoption of the technology. Here, we discuss the design of experiments necessary to achieve accurate prediction of GEBV in future generations in terms of the number of markers necessary and the size of the reference population where marker effects are estimated. We also present a simple method for implementing genomic selection using a genomic relationship matrix. Future challenges discussed include using whole genome sequence data to improve the accuracy of genomic selection and management of inbreeding through genomic relationships.     

Linkage disequilibrium in European elite maize germplasm investigated with SSRs

Information about the extent and genomic distribution of linkage disequilibrium (LD) is of fundamental importance for association mapping. The main objectives of this study were to (1) investigate genetic diversity within germplasm groups of elite European maize (Zea mays L.) inbred lines, (2) examine the population structure of elite European maize germplasm, and (3) determine the extent and genomic distribution of LD between pairs of simple sequence repeat (SSR) markers. We examined genetic diversity and LD in a cross section of European and US elite breeding material comprising 147 inbred lines genotyped with 100 SSR markers. For gene diversity within each group, significant (P<0.05) differences existed among the groups. The LD was significant (P<0.05) for 49% of the SSR marker pairs in the 80 flint lines and for 56% of the SSR marker pairs in the 57 dent lines. The ratio of linked to unlinked loci in LD was 1.1 for both germplasm groups. The high incidence of LD suggests that the extent of LD between SSR markers should allow the detection of marker-phenotype associations in a genome scan. However, our results also indicate that a high proportion of the observed LD is generated by forces, such as relatedness, population stratification, and genetic drift, which cause a high risk of detecting false positives in association mapping.     

基因组选择在猪杂交育种中的应用

基因组选择是指在全基因组范围内通过基因组中大量的标记信息估计出个体全基因组范围的育种值,可进一步提升育种效率和准确性,目前在猪纯繁育种中得到广泛应用。但有研究表明,现有的基因组选择方法在猪杂交育种上的应用效果并不理想,在跨群体条件下预测准确性极低。杂交作为养猪业中最为广泛的育种手段之一,通过结合基因组选择理论进一步提升猪的生产性能,具有重要的经济和研究价值。本文综述了基因组选择的发展及其在猪育种中的应用现状,并结合国内外猪杂交育种的方式,分析了目前基因组选择方法在猪杂交育种应用方面的不足,旨在为未来基因组选择在猪杂交育种中的合理应用提供参考。     

Genetic drift and selection effects of modified recurrent full-sib selection programs in two F2 populations of European flint maize

Selection response of a modified recurrent full-sib (FS) selection scheme conducted in two European flint F₂ maize (Zea mays L.) populations was re-evaluated. Our objectives were to (1) determine the selection response for per se and testcross performance in both populations and (2) separate genetic effects due to selection from those due to random genetic drift. Modified recurrent FS selection was conducted at three locations using an effective population size N _e = 32 and a selection rate of 25% for a selection index, based on grain yield and grain moisture. Recombination was performed according to a pseudo-factorial mating scheme. Selection response was assessed using a population diallel including the source population and advanced selection cycles, as well as testcrosses with unrelatesd inbred line testers and the parental F₁ generation. Selection response per cycle was significant for grain yield and grain moisture in both populations. Effects of random genetic drift caused only a small reduction in the selection response. No significant selection response was observed for testcrosses, suggesting that for heterotic traits, such as grain yield, a high frequency of favorable alleles in the elite tester masked the effects of genes segregating in the populations. We conclude that our modified recurrent FS selection is an alternative to other commonly applied intrapopulation recurrent selection schemes, and some of its features may also be useful for increasing the efficiency of interpopulation recurrent selection programs.     

Potential of genotyping-by-sequencing for genomic selection in livestock populations

Background

Next-generation sequencing techniques, such as genotyping-by-sequencing (GBS), provide alternatives to single nucleotide polymorphism (SNP) arrays. The aim of this work was to evaluate the potential of GBS compared to SNP array genotyping for genomic selection in livestock populations.

Methods

The value of GBS was quantified by simulation analyses in which three parameters were varied: (i) genome-wide sequence read depth (x) per individual from 0.01x to 20x or using SNP array genotyping; (ii) number of genotyped markers from 3000 to 300 000; and (iii) size of training and prediction sets from 500 to 50 000 individuals. The latter was achieved by distributing the total available x of 1000x, 5000x, or 10 000x per genotyped locus among the varying number of individuals. With SNP arrays, genotypes were called from sequence data directly. With GBS, genotypes were called from sequence reads that varied between loci and individuals according to a Poisson distribution with mean equal to x. Simulated data were analyzed with ridge regression and the accuracy and bias of genomic predictions and response to selection were quantified under the different scenarios.

Results

Accuracies of genomic predictions using GBS data or SNP array data were comparable when large numbers of markers were used and x per individual was ~1x or higher. The bias of genomic predictions was very high at a very low x. When the total available x was distributed among the training individuals, the accuracy of prediction was maximized when a large number of individuals was used that had GBS data with low x for a large number of markers. Similarly, response to selection was maximized under the same conditions due to increasing both accuracy and selection intensity.

Conclusions

GBS offers great potential for developing genomic selection in livestock populations because it makes it possible to cover large fractions of the genome and to vary the sequence read depth per individual. Thus, the accuracy of predictions is improved by increasing the size of training populations and the intensity of selection is increased by genotyping a larger number of selection candidates.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-015-0102-z) contains supplementary material, which is available to authorized users.     

Size of breeding populations required for selection programs

The minimum population size required for selection in order to reduce the effect of genetic drift to a particular level has been considered. The model of Nicholas was extended to include the measurement-error variance in the response variance. Situations where the sex ratios among scored and breeding individuals are unequal are also considered. When the duration of a selection experiment is relatively long, Nicholas' approximation (i.e., assuming that measurement error is negligible relative to drift) is useful in determining the minimum effective population size required. However, the measurement-error variance becomes an important source of variation in short-term ( 5 generations) selection experiments, and should not be ignored.     

A low-marker density implementation of genomic selection in aquaculture using within-family genomic breeding values

Background

Genomic selection can increase genetic gain within aquaculture breeding programs, but the high costs related to high-density genotyping of a large number of individuals would make the breeding program expensive. In this study, a low-cost method using low-density genotyping of pre-selected candidates and their sibs was evaluated by stochastic simulation.

Methods

A breeding scheme with selection for two traits, one measured on candidates and one on sibs was simulated. Genomic breeding values were estimated within families and combined with conventional family breeding values for candidates that were pre-selected based on conventional BLUP breeding values. This strategy was compared with a conventional breeding scheme and a full genomic selection program for which genomic breeding values were estimated across the whole population. The effects of marker density, level of pre-selection and number of sibs tested and genotyped for the sib-trait were studied.

Results

Within-family genomic breeding values increased genetic gain by 15% and reduced rate of inbreeding by 15%. Genetic gain was robust to a reduction in marker density, with only moderate reductions, even for very low densities. Pre-selection of candidates down to approximately 10% of the candidates before genotyping also had minor effects on genetic gain, but depended somewhat on marker density. The number of test-individuals, i.e. individuals tested for the sib-trait, affected genetic gain, but the fraction of the test-individuals genotyped only affected the relative contribution of each trait to genetic gain.

Conclusions

A combination of genomic within-family breeding values, based on low-density genotyping, and conventional BLUP family breeding values was shown to be a possible low marker density implementation of genomic selection for species with large full-sib families for which the costs of genotyping must be kept low without compromising the effect of genomic selection on genetic gain.     

Efficiency of genomic selection in an established commercial layer breeding program

Background

In breeding programs for layers, selection of hens and cocks is based on recording phenotypic data from hens in different housing systems. Genomic information can provide additional information for selection and/or allow for a strong reduction in the generation interval. In this study, a typical conventional layer breeding program using a four-line cross was modeled and the expected genetic progress was derived deterministically with the software ZPLAN+. This non-genomic reference scenario was compared to two genomic breeding programs to determine the best strategy for implementing genomic information in layer breeding programs.

Results

In scenario I, genomic information was used in addition to all other information available in the conventional breeding program, so the generation interval was the same as in the reference scenario, i.e. 14.5 months. Here, we assumed that either only young cocks or young cocks and hens were genotyped as selection candidates. In scenario II, we assumed that breeders of both sexes were used at the biologically earliest possible age, so that at the time of selection only performance data of the parent generation and genomic information of the selection candidates were available. In this case, the generation interval was reduced to eight months. In both scenarios, the number of genotyped male selection candidates was varied between 800 and 4800 males and two sizes of the calibration set (500 or 2000 animals) were considered. All genomic scenarios increased the expected genetic gain and the economic profit of the breeding program. In scenario II, the increase was much more pronounced and even in the most conservative implementation led to a 60% improvement in genetic gain and economic profit. This increase was in all cases associated with higher breeding costs.

Conclusions

While genomic selection is shown to have the potential to improve genetic gain in layer breeding programs, its implementation remains a business decision of the breeding company; the possible extra profit for the breeding company depends on whether the customers of breeding stock are willing to pay more for improved genetic quality.     

Temporal changes in allele frequencies in two European F2 flint maize populations under modified recurrent full-sib selection

Selection and random genetic drift are the two main forces affecting the selection response of recurrent selection (RS) programs by changes in allele frequencies. Therefore, detailed knowledge on allele frequency changes attributable to these forces is of fundamental importance for assessing RS programs. The objectives of our study were to (1) estimate the number, position, and genetic effect of quantitative trait loci (QTL) for selection index and its components in the base populations, (2) determine changes in allele frequencies of QTL regions due to the effects of random genetic drift and selection, and (3) predict allele frequency changes by using QTL results and compare these predictions with observed values. We performed QTL analyses, based on restriction fragment length polymorphisms (RFLPs) and simple sequence repeats (SSRs), in 274 F_2:3 lines of cross KW1265 × D146 (A × B) and 133 F_3:4 lines of cross D145 × KW1292 (C × D) originating from two European flint maize populations. Four (A × B) and seven (C × D) cycles of RS were analyzed with SSRs for significant allele frequency changes due to selection. Several QTL regions for selection index were detected with simple and composite interval mapping. In some of them, flanking markers showed a significant allele frequency change after the first and the final selection cycles. The correlation between observed and predicted allele frequencies was significant only in A × B. We attribute these observations mainly to (1) the high dependence of the power of QTL detection on the population size and (2) the occurrence of undetectable QTL in repulsion phase. Assessment of allele frequency changes in RS programs can be used to detect marker alleles linked to QTL regions under selection pressure. Electronic supplementary material The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Practical application of genomic selection in a doubled-haploid winter wheat breeding program

Crop improvement is a long-term, expensive institutional endeavor. Genomic selection (GS), which uses single nucleotide polymorphism (SNP) information to estimate genomic breeding values, has proven efficient to increasing genetic gain by accelerating the breeding process in animal breeding programs. As for crop improvement, with few exceptions, GS applicability remains in the evaluation of algorithm performance. In this study, we examined factors related to GS applicability in line development stage for grain yield using a hard red winter wheat (Triticum aestivum L.) doubled-haploid population. The performance of GS was evaluated in two consecutive years to predict grain yield. In general, the semi-parametric reproducing kernel Hilbert space prediction algorithm outperformed parametric genomic best linear unbiased prediction. For both parametric and semi-parametric algorithms, an upward bias in predictability was apparent in within-year cross-validation, suggesting the prerequisite of cross-year validation for a more reliable prediction. Adjusting the training population’s phenotype for genotype by environment effect had a positive impact on GS model’s predictive ability. Possibly due to marker redundancy, a selected subset of SNPs at an absolute pairwise correlation coefficient threshold value of 0.4 produced comparable results and reduced the computational burden of considering the full SNP set. Finally, in the context of an ongoing breeding and selection effort, the present study has provided a measure of confidence based on the deviation of line selection from GS results, supporting the implementation of GS in wheat variety development.     

Optimizing the allocation of resources for genomic selection in one breeding cycle

Key message

We developed a universally applicable planning tool for optimizing the allocation of resources for one cycle of genomic selection in a biparental population. The framework combines selection theory with constraint numerical optimization and considers genotype? ×? environment interactions.

Abstract

Genomic selection (GS) is increasingly implemented in plant breeding programs to increase selection gain but little is known how to optimally allocate the resources under a given budget. We investigated this problem with model calculations by combining quantitative genetic selection theory with constraint numerical optimization. We assumed one selection cycle where both the training and prediction sets comprised double haploid (DH) lines from the same biparental population. Grain yield for testcrosses of maize DH lines was used as a model trait but all parameters can be adjusted in a freely available software implementation. An extension of the expected selection accuracy given by Daetwyler et al. (2008) was developed to correctly balance between the number of environments for phenotyping the training set and its population size in the presence of genotype?×?environment interactions. Under small budget, genotyping costs mainly determine whether GS is superior over phenotypic selection. With increasing budget, flexibility in resource allocation increases greatly but selection gain leveled off quickly requiring balancing the number of populations with the budget spent for each population. The use of an index combining phenotypic and GS predicted values in the training set was especially beneficial under limited resources and large genotype × environment interactions. Once a sufficiently high selection accuracy is achieved in the prediction set, further selection gain can be achieved most efficiently by massively expanding its size. Thus, with increasing budget, reducing the costs for producing a DH line becomes increasingly crucial for successfully exploiting the benefits of GS.     

Effect of genomic prediction on response to selection in forest tree breeding

Through stochastic simulations, estimates of breeding values accuracies and response to selection were assessed under traditional pedigree-based and genomic-based evaluation methods. More specifically, several key parameters such as the trait’s heritability (0.2 and 0.6), the number of QTLs underlying the trait (100 to 200), and the marker density (1 to 10 SNPs/cM) were evaluated. Additionally, impact of two contrasting mating designs (partial diallel vs. single-pair mating) was investigated. Response to selection was then assessed in a seed production population (seed orchard consisting of unrelated selections) for different effective population sizes (N_e?=?5 to 25). The simulated candidate population comprised a fixed size of 2050 individuals with fast linkage disequilibrium decay, generally found in forest tree populations. Following the genetic/genomic evaluation, top-ranked individuals were selected to meeting the predetermined effective population size in target production population. The combination of low h², high N_e, and dense marker coverage resulted at maximum relative genomic prediction efficiency and the most efficient exploitation of the Mendelian sampling term (within-family additive genetic variance). Since genomic prediction of breeding values constitutes the methodological foundation of genomic selection, our results can be used to address important questions when similar scenarios are considered.     

RFLP diversity and relationships among traditional European maize populations

Given the large extent of hybrid cultivation, the importance of conserving the diversity of crop genetic resources has given birth to numerous collections of old races. In the present paper, we conduct a molecular characterisation of a large collection of 488 European maize populations using the bulk RFLP analysis. The analysis of 23 RFLP loci showed a high allelic richness of 11.5 alleles per locus. Populations from eastern Europe (Poland, Austria, Germany, etc.) showed the lowest genetic diversity, a lower number of unique alleles and a higher percentage of fixed loci than populations from southern Europe. In fact, genetic diversity appeared higher in Southern regions where the first maize populations are thought to have been introduced. Molecular classification based on Rogers' distance (i.e. alleles frequencies) allowed us to distinguish three main clusters which were highly consistent with geographic origins. A Northeastern cluster grouped together early or intermediate populations from Northeastern countries and the Balkans, a southeastern cluster joined late and partially dent populations from Greece and Italy, and, a southwestern cluster was made up of early flint populations from northern Spain, Portugal and the Pyrenees. A correlation between allelic frequencies at some loci and latitude and/or longitude was observed. Such tendencies may reflect the direction of gene flow between different races of maize: for instance, North American (Northern flint) and Caribbean populations were introduced, respectively, to northern and southern Europe, in the past.     

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.     

Optimization of genomic selection training populations with a genetic algorithm

In this article, we imagine a breeding scenario with a population of individuals that have been genotyped but not phenotyped. We derived a computationally efficient statistic that uses this genetic information to measure the reliability of genomic estimated breeding values (GEBV) for a given set of individuals (test set) based on a training set of individuals. We used this reliability measure with a genetic algorithm scheme to find an optimized training set from a larger set of candidate individuals. This subset was phenotyped to create the training set that was used in a genomic selection model to estimate GEBV in the test set. Our results show that, compared to a random sample of the same size, the use of a set of individuals selected by our method improved accuracies. We implemented the proposed training selection methodology on four sets of data on Arabidopsis, wheat, rice and maize. This dynamic model building process that takes genotypes of the individuals in the test sample into account while selecting the training individuals improves the performance of genomic selection models.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-015-0116-6) contains supplementary material, which is available to authorized users.