Search

Abstract Background The accuracy of genomic prediction is highly dependent on the size of the reference population. For small populations, including information from other populations could improve this accuracy. The usual strategy is to pool data from different populations; however, this has not proven as successful as hoped for with distantly related breeds. BayesRS is a novel approach to share information across populations for genomic predictions. The approach allows information to be captured even where the phase of SNP alleles and casuative mutation alleles are reversed across populations, or the actual casuative mutation is different between the populations but affects the same gene. Proportions of a four-distribution mixture for SNP effects in segments of fixed size along the genome are derived from one population and set as location specific prior proportions of distributions of SNP effects for the target population. The model was tested using dairy cattle populations of different breeds: 540 Australian Jersey bulls, 2297 Australian Holstein bulls and 5214 Nordic Holstein bulls. The traits studied were protein-, fat- and milk yield. Genotypic data was Illumina 777K SNPs, real or imputed. Results Results showed an increase in accuracy of up to 3.5% for the Jersey population when using BayesRS with a prior derived from Australian Holstein compared to a model without location specific priors. The increase in accuracy was however lower than was achieved when reference populations were combined to estimate SNP effects, except in the case of fat yield. The small size of the Jersey validation set meant that these improvements in accuracy were not significant using a Hotelling-Williams t-test at the 5% level. An increase in accuracy of 1-2% for all traits was observed in the Australian Holstein population when using a prior derived from the Nordic Holstein population compared to using no prior information. These improvements were significant (P Conclusion For some traits the method might be advantageous compared to pooling of reference data for distantly related populations, but further investigation is needed to confirm the results. For closely related populations the method does not perform better than pooling reference data. However, it does give an increased accuracy compared to analysis based on only one reference population, without an increased computational burden. The approach described here provides a general setup for inclusion of location specific priors: the approach could be used to include biological information in genomic predictions.

Background: Genotyping and whole-genome sequencing data have been generated for hundreds of thousands of cattle. International consortia used these data to compile imputation reference panels that facilitate the imputation of sequence variant genotypes for animals that have been genotyped using dense microarrays. Association studies with imputed sequence variant genotypes allow for the characterization of quantitative trait loci (QTL) at nucleotide resolution particularly when individuals from several breeds are included in the mapping populations. Results: We imputed genotypes for 28 million sequence variants in 17,229 cattle of the Braunvieh, Fleckvieh and Holstein breeds in order to compile large mapping populations that provide high power to identify QTL for milk production traits. Association tests between imputed sequence variant genotypes and fat and protein percentages in milk uncovered between six and thirteen QTL (P < 1e-8) per breed. Eight of the detected QTL were significant in more than one breed. We combined the results across breeds using meta-analysis and identified a total of 25 QTL including six that were not significant in the within-breed association studies. Two missense mutations in the ABCG2 (p.Y581S, rs43702337, P = 4.3e-34) and GHR (p.F279Y, rs385640152, P = 1.6e-74) genes were the top variants at QTL on chromosomes 6 and 20. Another known causal missense mutation in the DGAT1 gene (p.A232K, rs109326954, P = 8.4e-1436) was the second top variant at a QTL on chromosome 14 but its allelic substitution effects were inconsistent across breeds. It turned out that the conflicting allelic substitution effects resulted from flaws in the imputed genotypes due to the use of a multi-breed reference population for genotype imputation. Conclusions: Many QTL for milk production traits segregate across breeds and across-breed meta-analysis has greater power to detect such QTL than within-breed association testing. Association testing between imputed sequence variant genotypes and phenotypes of interest facilitates identifying causal mutations provided the accuracy of imputation is high. However, true causal mutations may remain undetected when the imputed sequence variant genotypes contain flaws. It is highly recommended to validate the effect of known causal variants in order to assess the ability to detect true causal mutations in association studies with imputed sequence variants. [ABSTRACT FROM AUTHOR]

Background: Using whole genome sequence data might improve genomic prediction accuracy, when compared with high-density SNP arrays, and could lead to identification of casual mutations affecting complex traits. For some traits, the most accurate genomic predictions are achieved with non-linear Bayesian methods. However, as the number of variants and the size of the reference population increase, the computational time required to implement these Bayesian methods (typically with Monte Carlo Markov Chain sampling) becomes unfeasibly long. Results: Here, we applied a new method, HyB_BR (for Hybrid BayesR), which implements a mixture model of normal distributions and hybridizes an Expectation-Maximization (EM) algorithm followed by Markov Chain Monte Carlo (MCMC) sampling, to genomic prediction in a large dairy cattle population with imputed whole genome sequence data. The imputed whole genome sequence data included 994,019 variant genotypes of 16,214 Holstein and Jersey bulls and cows. Traits included fat yield, milk volume, protein kg, fat% and protein% in milk, as well as fertility and heat tolerance. HyB_BR achieved genomic prediction accuracies as high as the full MCMC implementation of BayesR, both for predicting a validation set of Holstein and Jersey bulls (multi-breed prediction) and a validation set of Australian Red bulls (across-breed prediction). HyB_BR had a ten fold reduction in compute time, compared with the MCMC implementation of BayesR (48 hours versus 594 hours). We also demonstrate that in many cases HyB_BR identified sequence variants with a high posterior probability of affecting the milk production or fertility traits that were similar to those identified in BayesR. For heat tolerance, both HyB_BR and BayesR found variants in or close to promising candidate genes associated with this trait and not detected by previous studies. Conclusions: The results demonstrate that HyB_BR is a feasible method for simultaneous genomic prediction and QTL mapping with whole genome sequence in large reference populations. [ABSTRACT FROM AUTHOR]

Background: Bayesian mixture models in which the effects of SNP are assumed to come from normal distributions with different variances are attractive for simultaneous genomic prediction and QTL mapping. These models are usually implemented with Monte Carlo Markov Chain (MCMC) sampling, which requires long compute times with large genomic data sets. Here, we present an efficient approach (termed HyB_BR), which is a hybrid of an Expectation-Maximisation algorithm, followed by a limited number of MCMC without the requirement for burn-in. Results: To test prediction accuracy from HyB_BR, dairy cattle and human disease trait data were used. In the dairy cattle data, there were four quantitative traits (milk volume, protein kg, fat% in milk and fertility) measured in 16,214 cattle from two breeds genotyped for 632,002 SNPs. Validation of genomic predictions was in a subset of cattle either from the reference set or in animals from a third breeds that were not in the reference set. In all cases, HyB_BR gave almost identical accuracies to Bayesian mixture models implemented with full MCMC, however computational time was reduced by up to 1/17 of that required by full MCMC. The SNPs with high posterior probability of a non-zero effect were also very similar between full MCMC and HyB_BR, with several known genes affecting milk production in this category, as well as some novel genes. HyB_BR was also applied to seven human diseases with 4890 individuals genotyped for around 300 K SNPs in a case/control design, from the Welcome Trust Case Control Consortium (WTCCC). In this data set, the results demonstrated again that HyB_BR performed as well as Bayesian mixture models with full MCMC for genomic predictions and genetic architecture inference while reducing the computational time from 45 h with full MCMC to 3 h with HyB_BR. Conclusions: The results for quantitative traits in cattle and disease in humans demonstrate that HyB_BR can perform equally well as Bayesian mixture models implemented with full MCMC in terms of prediction accuracy, but with up to 17 times faster than the full MCMC implementations. The HyB_BR algorithm makes simultaneous genomic prediction, QTL mapping and inference of genetic architecture feasible in large genomic data sets. [ABSTRACT FROM AUTHOR]

Background: Copy number variants (CNVs) are a type of polymorphism found to underlie phenotypic variation, both in humans and livestock. Most surveys of CNV in livestock have been conducted in the cattle genome, and often utilise only a single approach for the detection of copy number differences. Here we performed a study of CNV in sheep, using multiple methods to identify and characterise copy number changes. Comprehensive information from small pedigrees (trios) was collected using multiple platforms (array CGH, SNP chip and whole genome sequence data), with these data then analysed via multiple approaches to identify and verify CNVs. Results: In total, 3,488 autosomal CNV regions (CNVRs) were identified in this study, which substantially builds on an initial survey of the sheep genome that identified 135 CNVRs. The average length of the identified CNVRs was 19 kb (range of 1 kb to 3.6 Mb), with shorter CNVRs being more frequent than longer CNVRs. The total length of all CNVRs was 67.6Mbps, which equates to 2.7 % of the sheep autosomes. For individuals this value ranged from 0.24 to 0.55 %, and the majority of CNVRs were identified in single animals. Rather than being uniformly distributed throughout the genome, CNVRs tended to be clustered. Application of three independent approaches for CNVR detection facilitated a comparison of validation rates. CNVs identified on the Roche-NimbleGen 2.1M CGH array generally had low validation rates with lower density arrays, while whole genome sequence data had the highest validation rate (>60 %). Conclusions: This study represents the first comprehensive survey of the distribution, prevalence and characteristics of CNVR in sheep. Multiple approaches were used to detect CNV regions and it appears that the best method for verifying CNVR on a large scale involves using a combination of detection methodologies. The characteristics of the 3,488 autosomal CNV regions identified in this study are comparable to other CNV regions reported in the literature and provide a valuable and sizeable addition to the small subset of published sheep CNVs. [ABSTRACT FROM AUTHOR]

Background: Genetic variation in both the composition and distribution of fat and muscle in the body is important to human health as well as the healthiness and value of meat from cattle and sheep. Here we use detailed phenotyping and a multi-trait approach to identify genes explaining variation in body composition traits. Results: A multi-trait genome wide association analysis of 56 carcass composition traits measured on 10,613 sheep with imputed and real genotypes on 510,174 SNPs was performed. We clustered 71 significant SNPs into five groups based on their pleiotropic effects across the 56 traits. Among these 71 significant SNPs, one group of 11 SNPs affected the fatty acid profile of themuscle and were close to 8 genes involved in fatty acid or triglyceride synthesis. Another group of 23 SNPs had an effect on mature size, based on their pattern of effects across traits, but the genes near this group of SNPs did not share any obvious function. Many of the likely candidate genes near SNPs with significant pleiotropic effects on the 56 traits are involved in intra-cellular signalling pathways. Among the significant SNPs were some with a convincing candidate gene due to the function of the gene (e.g. glycogen synthase affecting glycogen concentration) or because the same gene was associated with similar traits in other species. Conclusions: Using a multi-trait analysis increased the power to detect associations between SNP and body composition traits compared with the single trait analyses. Detailed phenotypic information helped to identify a convincing candidate in some cases as did information from other species. [ABSTRACT FROM AUTHOR]

Background: Allele specific gene expression (ASE), with the paternal allele more expressed than the maternal allele or vice versa, appears to be a common phenomenon in humans and mice. In other species the extent of ASE is unknown, and even in humans and mice there are several outstanding questions. These include; to what extent is ASE tissue specific? how often does the direction of allele expression imbalance reverse between tissues? how often is only one of the two alleles expressed? is there a genome wide bias towards expression of the paternal or maternal allele; and finally do genes that are nearby on a chromosome share the same direction of ASE? Here we use gene expression data (RNASeq) from 18 tissues from a single cow to investigate each of these questions in turn, and then validate some of these findings in two tissues from 20 cows. Results: Between 40 and 100 million sequence reads were generated per tissue across three replicate samples for each of the eighteen tissues from the single cow (the discovery dataset). A bovine gene expression atlas was created (the first from RNASeq data), and differentially expressed genes in each tissue were identified. To analyse ASE, we had access to unambiguously phased genotypes for all heterozygous variants in the cow's whole genome sequence, where these variants were homozygous in the whole genome sequence of her sire, and as a result we were able to map reads to parental genomes, to determine SNP and genes showing ASE in each tissue. In total 25,251 heterozygous SNP within 7985 genes were tested for ASE in at least one tissue. ASE was pervasive, 89 % of genes tested had significant ASE in at least one tissue. This large proportion of genes displaying ASE was confirmed in the two tissues in a validation dataset. For individual tissues the proportion of genes showing significant ASE varied from as low as 8-16 % of those tested in thymus to as high as 71-82 % of those tested in lung. There were a number of cases where the direction of allele expression imbalance reversed between tissues. For example the gene SPTY2D1 showed almost complete paternal allele expression in kidney and thymus, and almost complete maternal allele expression in the brain caudal lobe and brain cerebellum. Mono allelic expression (MAE) was common, with 1349 of 4856 genes (28 %) tested with more than one heterozygous SNP showing MAE. Across all tissues, 54.17 % of all genes with ASE favoured the paternal allele. Genes that are closely linked on the chromosome were more likely to show higher expression of the same allele (paternal or maternal) than expected by chance. We identified several long runs of neighbouring genes that showed either paternal or maternal ASE, one example was five adjacent genes (GIMAP8, GIMAP7 copy1, GIMAP4, GIMAP7 copy 2 and GIMAP5) that showed almost exclusive paternal expression in brain caudal lobe. Conclusions: Investigating the extent of ASE across 18 bovine tissues in one cow and two tissues in 20 cows demonstrated 1) ASE is pervasive in cattle, 2) the ASE is often MAE but ranges from MAE to slight overexpression of the major allele, 3) the ASE is most often tissue specific and that more than half the time displays divergent allele specific expression patterns across tissues, 4) across all genes there is a slight bias towards expression of the paternal allele and 5) genes expressing the same parental allele are clustered together more than expected by chance, and there are several runs of large numbers of genes expressing the same parental allele. [ABSTRACT FROM AUTHOR]

Background: Selection signatures aim to identify genomic regions underlying recent adaptations in populations. However, the effects of selection in the genome are difficult to distinguish from random processes, such as genetic drift. Often associations between selection signatures and selected variants for complex traits is assumed even though this is rarely (if ever) tested. In this paper, we use 8 breeds of domestic cattle under strong artificial selection to investigate if selection signatures are co-located in genomic regions which are likely to be under selection. Results: Our approaches to identify selection signatures (haplotype heterozygosity, integrated haplotype score and FST) identified strong and recent selection near many loci with mutations affecting simple traits under strong selection, such as coat colour. However, there was little evidence for a genome-wide association between strong selection signatures and regions affecting complex traits under selection, such as milk yield in dairy cattle. Even identifying selection signatures near some major loci was hindered by factors including allelic heterogeneity, selection for ancestral alleles and interactions with nearby selected loci. Conclusions: Selection signatures detect loci with large effects under strong selection. However, the methodology is often assumed to also detect loci affecting complex traits where the selection pressure at an individual locus is weak. We present empirical evidence to suggests little discernible ‘selection signature’ for complex traits in the genome of dairy cattle despite very strong and recent artificial selection. [ABSTRACT FROM AUTHOR]

Background: The accuracy of genomic prediction is highly dependent on the size of the reference population. For small populations, including information from other populations could improve this accuracy. The usual strategy is to pool data from different populations; however, this has not proven as successful as hoped for with distantly related breeds. BayesRS is a novel approach to share information across populations for genomic predictions. The approach allows information to be captured even where the phase of SNP alleles and casuative mutation alleles are reversed across populations, or the actual casuative mutation is different between the populations but affects the same gene. Proportions of a four-distribution mixture for SNP effects in segments of fixed size along the genome are derived from one population and set as location specific prior proportions of distributions of SNP effects for the target population. The model was tested using dairy cattle populations of different breeds: 540 Australian Jersey bulls, 2297 Australian Holstein bulls and 5214 Nordic Holstein bulls. The traits studied were protein-, fat- and milk yield. Genotypic data was Illumina 777K SNPs, real or imputed. Results: Results showed an increase in accuracy of up to 3.5% for the Jersey population when using BayesRS with a prior derived from Australian Holstein compared to a model without location specific priors. The increase in accuracy was however lower than was achieved when reference populations were combined to estimate SNP effects, except in the case of fat yield. The small size of the Jersey validation set meant that these improvements in accuracy were not significant using a Hotelling-Williams t-test at the 5% level. An increase in accuracy of 1-2% for all traits was observed in the Australian Holstein population when using a prior derived from the Nordic Holstein population compared to using no prior information. These improvements were significant (P<0.05) using the Hotelling Williams t-test for protein- and fat yield. Conclusion: For some traits the method might be advantageous compared to pooling of reference data for distantly related populations, but further investigation is needed to confirm the results. For closely related populations the method does not perform better than pooling reference data. However, it does give an increased accuracy compared to analysis based on only one reference population, without an increased computational burden. The approach described here provides a general setup for inclusion of location specific priors: the approach could be used to include biological information in genomic predictions. [ABSTRACT FROM AUTHOR]

Background: Using whole genome sequence data might improve genomic prediction accuracy, when compared with high-density SNP arrays, and could lead to identification of casual mutations affecting complex traits. For some traits, the most accurate genomic predictions are achieved with non-linear Bayesian methods. However, as the number of variants and the size of the reference population increase, the computational time required to implement these Bayesian methods (typically with Monte Carlo Markov Chain sampling) becomes unfeasibly long., Results: Here, we applied a new method, HyB&#95;BR (for Hybrid BayesR), which implements a mixture model of normal distributions and hybridizes an Expectation-Maximization (EM) algorithm followed by Markov Chain Monte Carlo (MCMC) sampling, to genomic prediction in a large dairy cattle population with imputed whole genome sequence data. The imputed whole genome sequence data included 994,019 variant genotypes of 16,214 Holstein and Jersey bulls and cows. Traits included fat yield, milk volume, protein kg, fat% and protein% in milk, as well as fertility and heat tolerance. HyB&#95;BR achieved genomic prediction accuracies as high as the full MCMC implementation of BayesR, both for predicting a validation set of Holstein and Jersey bulls (multi-breed prediction) and a validation set of Australian Red bulls (across-breed prediction). HyB&#95;BR had a ten fold reduction in compute time, compared with the MCMC implementation of BayesR (48 hours versus 594 hours). We also demonstrate that in many cases HyB&#95;BR identified sequence variants with a high posterior probability of affecting the milk production or fertility traits that were similar to those identified in BayesR. For heat tolerance, both HyB&#95;BR and BayesR found variants in or close to promising candidate genes associated with this trait and not detected by previous studies., Conclusions: The results demonstrate that HyB&#95;BR is a feasible method for simultaneous genomic prediction and QTL mapping with whole genome sequence in large reference populations.