The effect of high-density genotypic data and different methods on joint genomic prediction: A case study in large white pigs

Joint genomic prediction (GP) is an attractive method to improve the accuracy of GP by combining information from multiple populations. However, many factors can negatively influence the accuracy of joint GP, such as differences in linkage disequilibrium phasing between single nucleotide polymorphisms (SNPs) and causal variants, minor allele frequencies and causal variants’ effect sizes across different populations. The objective of this study was to investigate whether the imputed high-density genotype data can improve the accuracy of joint GP using genomic best linear unbiased prediction (GBLUP), single-step GBLUP (ssGBLUP), multi-trait GBLUP (MT-GBLUP) and GBLUP based on genomic relationship matrix considering heterogenous minor allele frequencies across different populations (wGBLUP). Three traits, including days taken to reach slaughter weight, backfat thickness and loin muscle area, were measured on 67 276 Large White pigs from two different populations, for which 3334 were genotyped by SNP array. The results showed that a combined population could substantially improve the accuracy of GP compared with a single-population GP, especially for the population with a smaller size. The imputed SNP data had no effect for single population GP but helped to yield higher accuracy than the medium-density array data for joint GP. Of the four methods, ssGLBUP performed the best, but the advantage of ssGBLUP decreased as more individuals were genotyped. In some cases, MT-GBLUP and wGBLUP performed better than GBLUP. In conclusion, our results confirmed that joint GP could be beneficial from imputed high-density genotype data, and the wGBLUP and MT-GBLUP methods are promising for joint GP in pig breeding.     

Genomic prediction using imputed whole-genome sequence data in Holstein Friesian cattle

Background

In contrast to currently used single nucleotide polymorphism (SNP) panels, the use of whole-genome sequence data is expected to enable the direct estimation of the effects of causal mutations on a given trait. This could lead to higher reliabilities of genomic predictions compared to those based on SNP genotypes. Also, at each generation of selection, recombination events between a SNP and a mutation can cause decay in reliability of genomic predictions based on markers rather than on the causal variants. Our objective was to investigate the use of imputed whole-genome sequence genotypes versus high-density SNP genotypes on (the persistency of) the reliability of genomic predictions using real cattle data.

Methods

Highly accurate phenotypes based on daughter performance and Illumina BovineHD Beadchip genotypes were available for 5503 Holstein Friesian bulls. The BovineHD genotypes (631,428 SNPs) of each bull were used to impute whole-genome sequence genotypes (12,590,056 SNPs) using the Beagle software. Imputation was done using a multi-breed reference panel of 429 sequenced individuals. Genomic estimated breeding values for three traits were predicted using a Bayesian stochastic search variable selection (BSSVS) model and a genome-enabled best linear unbiased prediction model (GBLUP). Reliabilities of predictions were based on 2087 validation bulls, while the other 3416 bulls were used for training.

Results

Prediction reliabilities ranged from 0.37 to 0.52. BSSVS performed better than GBLUP in all cases. Reliabilities of genomic predictions were slightly lower with imputed sequence data than with BovineHD chip data. Also, the reliabilities tended to be lower for both sequence data and BovineHD chip data when relationships between training animals were low. No increase in persistency of prediction reliability using imputed sequence data was observed.

Conclusions

Compared to BovineHD genotype data, using imputed sequence data for genomic prediction produced no advantage. To investigate the putative advantage of genomic prediction using (imputed) sequence data, a training set with a larger number of individuals that are distantly related to each other and genomic prediction models that incorporate biological information on the SNPs or that apply stricter SNP pre-selection should be considered.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-015-0149-x) contains supplementary material, which is available to authorized users.     

Impact of QTL properties on the accuracy of multi-breed genomic prediction

Background

Although simulation studies show that combining multiple breeds in one reference population increases accuracy of genomic prediction, this is not always confirmed in empirical studies. This discrepancy might be due to the assumptions on quantitative trait loci (QTL) properties applied in simulation studies, including number of QTL, spectrum of QTL allele frequencies across breeds, and distribution of allele substitution effects. We investigated the effects of QTL properties and of including a random across- and within-breed animal effect in a genomic best linear unbiased prediction (GBLUP) model on accuracy of multi-breed genomic prediction using genotypes of Holstein-Friesian and Jersey cows.

Methods

Genotypes of three classes of variants obtained from whole-genome sequence data, with moderately low, very low or extremely low average minor allele frequencies (MAF), were imputed in 3000 Holstein-Friesian and 3000 Jersey cows that had real high-density genotypes. Phenotypes of traits controlled by QTL with different properties were simulated by sampling 100 or 1000 QTL from one class of variants and their allele substitution effects either randomly from a gamma distribution, or computed such that each QTL explained the same variance, i.e. rare alleles had a large effect. Genomic breeding values for 1000 selection candidates per breed were estimated using GBLUP modelsincluding a random across- and a within-breed animal effect.

Results

For all three classes of QTL allele frequency spectra, accuracies of genomic prediction were not affected by the addition of 2000 individuals of the other breed to a reference population of the same breed as the selection candidates. Accuracies of both single- and multi-breed genomic prediction decreased as MAF of QTL decreased, especially when rare alleles had a large effect. Accuracies of genomic prediction were similar for the models with and without a random within-breed animal effect, probably because of insufficient power to separate across- and within-breed animal effects.

Conclusions

Accuracy of both single- and multi-breed genomic prediction depends on the properties of the QTL that underlie the trait. As QTL MAF decreased, accuracy decreased, especially when rare alleles had a large effect. This demonstrates that QTL properties are key parameters that determine the accuracy of genomic prediction.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-015-0124-6) contains supplementary material, which is available to authorized users.     

Pre-selecting markers based on fixation index scores improved the power of genomic evaluations in a combined Yorkshire pig population

Combining different swine populations in genomic prediction can be an important tool, leading to an increased accuracy of genomic prediction using single nucleotide polymorphism (SNP) chip data compared with within-population genomic. However, the expected higher accuracy of multi-population genomic prediction has not been realized. This may be due to an inconsistent linkage disequilibrium (LD) between SNPs and quantitative trait loci (QTL) across populations, and the weak genetic relationships across populations. In this study, we determined the impact of different genomic relationship matrices, SNP density and pre-selected variants on prediction accuracy using a combined Yorkshire pig population. Our objective was to provide useful strategies for improving the accuracy of genomic prediction within a combined population. Results showed that the accuracy of genomic best linear unbiased prediction (GBLUP) using imputed whole-genome sequencing (WGS) data in the combined population was always higher than that within populations. Furthermore, the use of imputed WGS data always resulted in a higher accuracy of GBLUP than the use of 80K chip data for the combined population. Additionally, the accuracy of GBLUP with a non-linear genomic relationship matrix was markedly increased (0.87% to 15.17% for 80K chip data, and 0.43% to 4.01% for imputed WGS data) compared with that obtained with a linear genomic relationship matrix, except for the prediction of XD population in the combined population using imputed WGS data. More importantly, the application of pre-selected variants based on fixation index (Fst) scores improved the accuracy of multi-population genomic prediction, especially for 80K chip data. For BLUP|GA (BLUP approach given the genetic architecture), the use of a linear method with an appropriate weight to build a weight-relatedness matrix led to a higher prediction accuracy compared with the use of only pre-selected SNPs for genomic evaluations, especially for the total number of piglets born. However, for the non-linear method, BLUP|GA showed only a small increase or even a decrease in prediction accuracy compared with the use of only pre-selected SNPs. Overall, the best genomic evaluation strategy for reproduction-related traits for a combined population was found to be GBLUP performed with a non-linear genomic relationship matrix using variants pre-selected from the 80K chip data based on Fst scores.     

Linkage disequilibrium among commonly genotyped SNP variants detected from bull sequence,

Genomic prediction utilizing causal variants could increase selection accuracy above that achieved with SNPs genotyped by currently available arrays used for genomic selection. A number of variants detected from sequencing influential sires are likely to be causal, but noticeable improvements in prediction accuracy using imputed sequence variant genotypes have not been reported. Improvement in accuracy of predicted breeding values may be limited by the accuracy of imputed sequence variants. Using genotypes of SNPs on a high‐density array and non‐synonymous SNPs detected in sequence from influential sires of a multibreed population, results of this examination suggest that linkage disequilibrium between non‐synonymous and array SNPs may be insufficient for accurate imputation from the array to sequence. In contrast to 75% of array SNPs being strongly correlated to another SNP on the array, less than 25% of the non‐synonymous SNPs were strongly correlated to an array SNP. When correlations between non‐synonymous and array SNPs were strong, distances between the SNPs were greater than separation that might be expected based on linkage disequilibrium decay. Consistently near‐perfect whole‐genome linkage disequilibrium between the full array and each non‐synonymous SNP within the sequenced bulls suggests that whole‐genome approaches to infer sequence variants might be more accurate than imputation based on local haplotypes. Opportunity for strong linkage disequilibrium between sequence and array SNPs may be limited by discrepancies in allele frequency distributions, so investigating alternate genotyping approaches and panels providing greater chances of frequency‐matched SNPs strongly correlated to sequence variants is also warranted. Genotypes used for this study are available from https://www.animalgenome.org/repository/pub/ ;USDA2017.0519/.     

The Effects of Demography and Long-Term Selection on the Accuracy of Genomic Prediction with Sequence Data

The use of dense SNPs to predict the genetic value of an individual for a complex trait is often referred to as “genomic selection” in livestock and crops, but is also relevant to human genetics to predict, for example, complex genetic disease risk. The accuracy of prediction depends on the strength of linkage disequilibrium (LD) between SNPs and causal mutations. If sequence data were used instead of dense SNPs, accuracy should increase because causal mutations are present, but demographic history and long-term negative selection also influence accuracy. We therefore evaluated genomic prediction, using simulated sequence in two contrasting populations: one reducing from an ancestrally large effective population size (N_e) to a small one, with high LD common in domestic livestock, while the second had a large constant-sized N_e with low LD similar to that in some human or outbred plant populations. There were two scenarios in each population; causal variants were either neutral or under long-term negative selection. For large N_e, sequence data led to a 22% increase in accuracy relative to ∼600K SNP chip data with a Bayesian analysis and a more modest advantage with a BLUP analysis. This advantage increased when causal variants were influenced by negative selection, and accuracy persisted when 10 generations separated reference and validation populations. However, in the reducing N_e population, there was little advantage for sequence even with negative selection. This study demonstrates the joint influence of demography and selection on accuracy of prediction and improves our understanding of how best to exploit sequence for genomic prediction.     

Accuracy of imputation to whole-genome sequence data in Holstein Friesian cattle

Background

The use of whole-genome sequence data can lead to higher accuracy in genome-wide association studies and genomic predictions. However, to benefit from whole-genome sequence data, a large dataset of sequenced individuals is needed. Imputation from SNP panels, such as the Illumina BovineSNP50 BeadChip and Illumina BovineHD BeadChip, to whole-genome sequence data is an attractive and less expensive approach to obtain whole-genome sequence genotypes for a large number of individuals than sequencing all individuals. Our objective was to investigate accuracy of imputation from lower density SNP panels to whole-genome sequence data in a typical dataset for cattle.

Methods

Whole-genome sequence data of chromosome 1 (1737 471 SNPs) for 114 Holstein Friesian bulls were used. Beagle software was used for imputation from the BovineSNP50 (3132 SNPs) and BovineHD (40 492 SNPs) beadchips. Accuracy was calculated as the correlation between observed and imputed genotypes and assessed by five-fold cross-validation. Three scenarios S40, S60 and S80 with respectively 40%, 60%, and 80% of the individuals as reference individuals were investigated.

Results

Mean accuracies of imputation per SNP from the BovineHD panel to sequence data and from the BovineSNP50 panel to sequence data for scenarios S40 and S80 ranged from 0.77 to 0.83 and from 0.37 to 0.46, respectively. Stepwise imputation from the BovineSNP50 to BovineHD panel and then to sequence data for scenario S40 improved accuracy per SNP to 0.65 but it varied considerably between SNPs.

Conclusions

Accuracy of imputation to whole-genome sequence data was generally high for imputation from the BovineHD beadchip, but was low from the BovineSNP50 beadchip. Stepwise imputation from the BovineSNP50 to the BovineHD beadchip and then to sequence data substantially improved accuracy of imputation. SNPs with a low minor allele frequency were more difficult to impute correctly and the reliability of imputation varied more. Linkage disequilibrium between an imputed SNP and the SNP on the lower density panel, minor allele frequency of the imputed SNP and size of the reference group affected imputation reliability.     

The impact of library preparation protocols on the consistency of allele frequency estimates in Pool‐Seq data

Sequencing pools of individuals (Pool‐Seq) is a cost‐effective method to determine genome‐wide allele frequency estimates. Given the importance of meta‐analyses combining data sets, we determined the influence of different genomic library preparation protocols on the consistency of allele frequency estimates. We found that typically no more than 1% of the variation in allele frequency estimates could be attributed to differences in library preparation. Also read length had only a minor effect on the consistency of allele frequency estimates. By far, the most pronounced influence could be attributed to sequence coverage. Increasing the coverage from 30‐ to 50‐fold improved the consistency of allele frequency estimates by at least 27%. We conclude that Pool‐Seq data can be easily combined across different library preparation methods, but sufficient sequence coverage is key to reliable results.     

Why are rare variants hard to impute? Coalescent models reveal theoretical limits in existing algorithms

Genotype imputation is an indispensable step in human genetic studies. Large reference panels with deeply sequenced genomes now allow interrogating variants with minor allele frequency < 1% without sequencing. Although it is critical to consider limits of this approach, imputation methods for rare variants have only done so empirically; the theoretical basis of their imputation accuracy has not been explored. To provide theoretical consideration of imputation accuracy under the current imputation framework, we develop a coalescent model of imputing rare variants, leveraging the joint genealogy of the sample to be imputed and reference individuals. We show that broadly used imputation algorithms include model misspecifications about this joint genealogy that limit the ability to correctly impute rare variants. We develop closed-form solutions for the probability distribution of this joint genealogy and quantify the inevitable error rate resulting from the model misspecification across a range of allele frequencies and reference sample sizes. We show that the probability of a falsely imputed minor allele decreases with reference sample size, but the proportion of falsely imputed minor alleles mostly depends on the allele count in the reference sample. We summarize the impact of this error on genotype imputation on association tests by calculating the r² between imputed and true genotype and show that even when modeling other sources of error, the impact of the model misspecification has a significant impact on the r² of rare variants. To evaluate these predictions in practice, we compare the imputation of the same dataset across imputation panels of different sizes. Although this empirical imputation accuracy is substantially lower than our theoretical prediction, modeling misspecification seems to further decrease imputation accuracy for variants with low allele counts in the reference. These results provide a framework for developing new imputation algorithms and for interpreting rare variant association analyses.     

Enlarging a training set for genomic selection by imputation of un-genotyped animals in populations of varying genetic architecture

Background

The most common application of imputation is to infer genotypes of a high-density panel of markers on animals that are genotyped for a low-density panel. However, the increase in accuracy of genomic predictions resulting from an increase in the number of markers tends to reach a plateau beyond a certain density. Another application of imputation is to increase the size of the training set with un-genotyped animals. This strategy can be particularly successful when a set of closely related individuals are genotyped.

Methods

Imputation on completely un-genotyped dams was performed using known genotypes from the sire of each dam, one offspring and the offspring’s sire. Two methods were applied based on either allele or haplotype frequencies to infer genotypes at ambiguous loci. Results of these methods and of two available software packages were compared. Quality of imputation under different population structures was assessed. The impact of using imputed dams to enlarge training sets on the accuracy of genomic predictions was evaluated for different populations, heritabilities and sizes of training sets.

Results

Imputation accuracy ranged from 0.52 to 0.93 depending on the population structure and the method used. The method that used allele frequencies performed better than the method based on haplotype frequencies. Accuracy of imputation was higher for populations with higher levels of linkage disequilibrium and with larger proportions of markers with more extreme allele frequencies. Inclusion of imputed dams in the training set increased the accuracy of genomic predictions. Gains in accuracy ranged from close to zero to 37.14%, depending on the simulated scenario. Generally, the larger the accuracy already obtained with the genotyped training set, the lower the increase in accuracy achieved by adding imputed dams.

Conclusions

Whenever a reference population resembling the family configuration considered here is available, imputation can be used to achieve an extra increase in accuracy of genomic predictions by enlarging the training set with completely un-genotyped dams. This strategy was shown to be particularly useful for populations with lower levels of linkage disequilibrium, for genomic selection on traits with low heritability, and for species or breeds for which the size of the reference population is limited.     

Design of a low‐density SNP chip for the main Australian sheep breeds and its effect on imputation and genomic prediction accuracy

Genotyping sheep for genome‐wide SNPs at lower density and imputing to a higher density would enable cost‐effective implementation of genomic selection, provided imputation was accurate enough. Here, we describe the design of a low‐density (12k) SNP chip and evaluate the accuracy of imputation from the 12k SNP genotypes to 50k SNP genotypes in the major Australian sheep breeds. In addition, the impact of imperfect imputation on genomic predictions was evaluated by comparing the accuracy of genomic predictions for 15 novel meat traits including carcass and meat quality and omega fatty acid traits in sheep, from 12k SNP genotypes, imputed 50k SNP genotypes and real 50k SNP genotypes. The 12k chip design included 12 223 SNPs with a high minor allele frequency that were selected with intermarker spacing of 50–475 kb. SNPs for parentage and horned or polled tests also were represented. Chromosome ends were enriched with SNPs to reduce edge effects on imputation. The imputation performance of the 12k SNP chip was evaluated using 50k SNP genotypes of 4642 animals from six breeds in three different scenarios: (1) within breed, (2) single breed from multibreed reference and (3) multibreed from a single‐breed reference. The highest imputation accuracies were found with scenario 2, whereas scenario 3 was the worst, as expected. Using scenario 2, the average imputation accuracy in Border Leicester, Polled Dorset, Merino, White Suffolk and crosses was 0.95, 0.95, 0.92, 0.91 and 0.93 respectively. Imputation scenario 2 was used to impute 50k genotypes for 10 396 animals with novel meat trait phenotypes to compare genomic prediction accuracy using genomic best linear unbiased prediction (GBLUP) with real and imputed 50k genotypes. The weighted mean imputation accuracy achieved was 0.92. The average accuracy of genomic estimated breeding values (GEBVs) based on only 12k data was 0.08 across traits and breeds, but accuracies varied widely. The mean GBLUP accuracies with imputed 50k data more than doubled to 0.21. Accuracies of genomic prediction were very similar for imputed and real 50k genotypes. There was no apparent impact on accuracy of GEBVs as a result of using imputed rather than real 50k genotypes, provided imputation accuracy was >90%.     

Accuracy of genotype imputation in sheep breeds

Although genomic selection offers the prospect of improving the rate of genetic gain in meat, wool and dairy sheep breeding programs, the key constraint is likely to be the cost of genotyping. Potentially, this constraint can be overcome by genotyping selection candidates for a low density (low cost) panel of SNPs with sparse genotype coverage, imputing a much higher density of SNP genotypes using a densely genotyped reference population. These imputed genotypes would then be used with a prediction equation to produce genomic estimated breeding values. In the future, it may also be desirable to impute very dense marker genotypes or even whole genome re‐sequence data from moderate density SNP panels. Such a strategy could lead to an accurate prediction of genomic estimated breeding values across breeds, for example. We used genotypes from 48 640 (50K) SNPs genotyped in four sheep breeds to investigate both the accuracy of imputation of the 50K SNPs from low density SNP panels, as well as prospects for imputing very dense or whole genome re‐sequence data from the 50K SNPs (by leaving out a small number of the 50K SNPs at random). Accuracy of imputation was low if the sparse panel had less than 5000 (5K) markers. Across breeds, it was clear that the accuracy of imputing from sparse marker panels to 50K was higher if the genetic diversity within a breed was lower, such that relationships among animals in that breed were higher. The accuracy of imputation from sparse genotypes to 50K genotypes was higher when the imputation was performed within breed rather than when pooling all the data, despite the fact that the pooled reference set was much larger. For Border Leicesters, Poll Dorsets and White Suffolks, 5K sparse genotypes were sufficient to impute 50K with 80% accuracy. For Merinos, the accuracy of imputing 50K from 5K was lower at 71%, despite a large number of animals with full genotypes (2215) being used as a reference. For all breeds, the relationship of individuals to the reference explained up to 64% of the variation in accuracy of imputation, demonstrating that accuracy of imputation can be increased if sires and other ancestors of the individuals to be imputed are included in the reference population. The accuracy of imputation could also be increased if pedigree information was available and was used in tracking inheritance of large chromosome segments within families. In our study, we only considered methods of imputation based on population‐wide linkage disequilibrium (largely because the pedigree for some of the populations was incomplete). Finally, in the scenarios designed to mimic imputation of high density or whole genome re‐sequence data from the 50K panel, the accuracy of imputation was much higher (86–96%). This is promising, suggesting that in silico genome re‐sequencing is possible in sheep if a suitable pool of key ancestors is sequenced for each breed.     

Estimation of Allele Frequencies From High-Coverage Genome-Sequencing Projects

A new generation of high-throughput sequencing strategies will soon lead to the acquisition of high-coverage genomic profiles of hundreds to thousands of individuals within species, generating unprecedented levels of information on the frequencies of nucleotides segregating at individual sites. However, because these new technologies are error prone and yield uneven coverage of alleles in diploid individuals, they also introduce the need for novel methods for analyzing the raw read data. A maximum-likelihood method for the estimation of allele frequencies is developed, eliminating both the need to arbitrarily discard individuals with low coverage and the requirement for an extrinsic measure of the sequence error rate. The resultant estimates are nearly unbiased with asymptotically minimal sampling variance, thereby defining the limits to our ability to estimate population-genetic parameters and providing a logical basis for the optimal design of population-genomic surveys.     

Comparison of genotype imputation strategies using a combined reference panel for chicken population

Using whole-genome sequence (WGS) data are supposed to be optimal for genome-wide association studies and genomic predictions. However, sequencing thousands of individuals of interest is expensive. Imputation from single nucleotide polymorphisms panels to WGS data is an attractive approach to obtain highly reliable WGS data at low cost. Here, we conducted a genotype imputation study with a combined reference panel in yellow-feather dwarf broiler population. The combined reference panel was assembled by sequencing 24 key individuals of a yellow-feather dwarf broiler population (internal reference panel) and WGS data from 311 chickens in public databases (external reference panel). Three scenarios were investigated to determine how different factors affect the accuracy of imputation from 600 K array data to WGS data, including: genotype imputation with internal, external and combined reference panels; the number of internal reference individuals in the combined reference panel; and different reference sizes and selection strategies of an external reference panel. Results showed that imputation accuracy from 600 K to WGS data were 0.834±0.012, 0.920±0.007 and 0.982±0.003 for the internal, external and combined reference panels, respectively. Increasing the reference size from 50 to 250 improved the accuracy of genotype imputation from 0.848 to 0.974 for the combined reference panel and from 0.647 to 0.917 for the external reference panel. The selection strategies for the external reference panel had no impact on the accuracy of imputation using the combined reference panel. However, if only an external reference panel with reference size >50 was used, the selection strategy of minimizing the average distance to the closest leaf had the greatest imputation accuracy compared with other methods. Generally, using a combined reference panel provided greater imputation accuracy, especially for low-frequency variants. In conclusion, the optimal imputation strategy with a combined reference panel should comprehensively consider genetic diversity of the study population, availability and properties of external reference panels, sequencing and computing costs, and frequency of imputed variants. This work sheds light on how to design and execute genotype imputation with a combined external reference panel in a livestock population.     

Evaluation of measures of correctness of genotype imputation in the context of genomic prediction: a review of livestock applications

In livestock, many studies have reported the results of imputation to 50k single nucleotide polymorphism (SNP) genotypes for animals that are genotyped with low-density SNP panels. The objective of this paper is to review different measures of correctness of imputation, and to evaluate their utility depending on the purpose of the imputed genotypes. Across studies, imputation accuracy, computed as the correlation between true and imputed genotypes, and imputation error rates, that counts the number of incorrectly imputed alleles, are commonly used measures of imputation correctness. Based on the nature of both measures and results reported in the literature, imputation accuracy appears to be a more useful measure of the correctness of imputation than imputation error rates, because imputation accuracy does not depend on minor allele frequency (MAF), whereas imputation error rate depends on MAF. Therefore imputation accuracy can be better compared across loci with different MAF. Imputation accuracy depends on the ability of identifying the correct haplotype of a SNP, but many other factors have been identified as well, including the number of genotyped immediate ancestors, the number of animals with genotypes at the high-density panel, the SNP density on the low- and high-density panel, the MAF of the imputed SNP and whether imputed SNP are located at the end of a chromosome or not. Some of these factors directly contribute to the linkage disequilibrium between imputed SNP and SNP on the low-density panel. When imputation accuracy is assessed as a predictor for the accuracy of subsequent genomic prediction, we recommend that: (1) individual-specific imputation accuracies should be used that are computed after centring and scaling both true and imputed genotypes; and (2) imputation of gene dosage is preferred over imputation of the most likely genotype, as this increases accuracy and reduces bias of the imputed genotypes and the subsequent genomic predictions.     

Comparing strategies for selection of low-density SNPs for imputation-mediated genomic prediction in U. S. Holsteins

SNP chips are commonly used for genotyping animals in genomic selection but strategies for selecting low-density (LD) SNPs for imputation-mediated genomic selection have not been addressed adequately. The main purpose of the present study was to compare the performance of eight LD (6K) SNP panels, each selected by a different strategy exploiting a combination of three major factors: evenly-spaced SNPs, increased minor allele frequencies, and SNP-trait associations either for single traits independently or for all the three traits jointly. The imputation accuracies from 6K to 80K SNP genotypes were between 96.2 and 98.2%. Genomic prediction accuracies obtained using imputed 80K genotypes were between 0.817 and 0.821 for daughter pregnancy rate, between 0.838 and 0.844 for fat yield, and between 0.850 and 0.863 for milk yield. The two SNP panels optimized on the three major factors had the highest genomic prediction accuracy (0.821–0.863), and these accuracies were very close to those obtained using observed 80K genotypes (0.825–0.868). Further exploration of the underlying relationships showed that genomic prediction accuracies did not respond linearly to imputation accuracies, but were significantly affected by genotype (imputation) errors of SNPs in association with the traits to be predicted. SNPs optimal for map coverage and MAF were favorable for obtaining accurate imputation of genotypes whereas trait-associated SNPs improved genomic prediction accuracies. Thus, optimal LD SNP panels were the ones that combined both strengths. The present results have practical implications on the design of LD SNP chips for imputation-enabled genomic prediction.     

Increasing calling accuracy,coverage, and read-depth in sequence data by the use of haplotype blocks

High-throughput genotyping of large numbers of lines remains a key challenge in plant genetics, requiring geneticists and breeders to find a balance between data quality and the number of genotyped lines under a variety of different existing genotyping technologies when resources are limited. In this work, we are proposing a new imputation pipeline (“HBimpute”) that can be used to generate high-quality genomic data from low read-depth whole-genome-sequence data. The key idea of the pipeline is the use of haplotype blocks from the software HaploBlocker to identify locally similar lines and subsequently use the reads of all locally similar lines in the variant calling for a specific line. The effectiveness of the pipeline is showcased on a dataset of 321 doubled haploid lines of a European maize landrace, which were sequenced at 0.5X read-depth. The overall imputing error rates are cut in half compared to state-of-the-art software like BEAGLE and STITCH, while the average read-depth is increased to 83X, thus enabling the calling of copy number variation. The usefulness of the obtained imputed data panel is further evaluated by comparing the performance of sequence data in common breeding applications to that of genomic data generated with a genotyping array. For both genome-wide association studies and genomic prediction, results are on par or even slightly better than results obtained with high-density array data (600k). In particular for genomic prediction, we observe slightly higher data quality for the sequence data compared to the 600k array in the form of higher prediction accuracies. This occurred specifically when reducing the data panel to the set of overlapping markers between sequence and array, indicating that sequencing data can benefit from the same marker ascertainment as used in the array process to increase the quality and usability of genomic data.     

Genome-Wide Prediction of Traits with Different Genetic Architecture Through Efficient Variable Selection

In genome-based prediction there is considerable uncertainty about the statistical model and method required to maximize prediction accuracy. For traits influenced by a small number of quantitative trait loci (QTL), predictions are expected to benefit from methods performing variable selection [e.g., BayesB or the least absolute shrinkage and selection operator (LASSO)] compared to methods distributing effects across the genome [ridge regression best linear unbiased prediction (RR-BLUP)]. We investigate the assumptions underlying successful variable selection by combining computer simulations with large-scale experimental data sets from rice (Oryza sativa L.), wheat (Triticum aestivum L.), and Arabidopsis thaliana (L.). We demonstrate that variable selection can be successful when the number of phenotyped individuals is much larger than the number of causal mutations contributing to the trait. We show that the sample size required for efficient variable selection increases dramatically with decreasing trait heritabilities and increasing extent of linkage disequilibrium (LD). We contrast and discuss contradictory results from simulation and experimental studies with respect to superiority of variable selection methods over RR-BLUP. Our results demonstrate that due to long-range LD, medium heritabilities, and small sample sizes, superiority of variable selection methods cannot be expected in plant breeding populations even for traits like FRIGIDA gene expression in Arabidopsis and flowering time in rice, assumed to be influenced by a few major QTL. We extend our conclusions to the analysis of whole-genome sequence data and infer upper bounds for the number of causal mutations which can be identified by LASSO. Our results have major impact on the choice of statistical method needed to make credible inferences about genetic architecture and prediction accuracy of complex traits.     

Accounting for genotype uncertainty in the estimation of allele frequencies in autopolyploids

Despite the increasing opportunity to collect large‐scale data sets for population genomic analyses, the use of high‐throughput sequencing to study populations of polyploids has seen little application. This is due in large part to problems associated with determining allele copy number in the genotypes of polyploid individuals (allelic dosage uncertainty–ADU), which complicates the calculation of important quantities such as allele frequencies. Here, we describe a statistical model to estimate biallelic SNP frequencies in a population of autopolyploids using high‐throughput sequencing data in the form of read counts. We bridge the gap from data collection (using restriction enzyme based techniques [e.g. GBS, RADseq]) to allele frequency estimation in a unified inferential framework using a hierarchical Bayesian model to sum over genotype uncertainty. Simulated data sets were generated under various conditions for tetraploid, hexaploid and octoploid populations to evaluate the model's performance and to help guide the collection of empirical data. We also provide an implementation of our model in the R package polyfreqs and demonstrate its use with two example analyses that investigate (i) levels of expected and observed heterozygosity and (ii) model adequacy. Our simulations show that the number of individuals sampled from a population has a greater impact on estimation error than sequencing coverage. The example analyses also show that our model and software can be used to make inferences beyond the estimation of allele frequencies for autopolyploids by providing assessments of model adequacy and estimates of heterozygosity.