Including microbiome information in a multi-trait genomic evaluation: a case study on longitudinal growth performance in beef cattle.

BACKGROUND: Growth rate is an important component of feed conversion efficiency in cattle and varies across the different stages of the finishing period. The metabolic effect of the rumen microbiome is essential for cattle growth, and investigating the genomic and microbial factors that underlie this temporal variation can help maximize feed conversion efficiency at each growth stage. RESULTS: By analysing longitudinal body weights during the finishing period and genomic and metagenomic data from 359 beef cattle, our study demonstrates that the influence of the host genome on the functional rumen microbiome contributes to the temporal variation in average daily gain (ADG) in different months (ADG1, ADG2, ADG3, ADG4). Five hundred and thirty-three additive log-ratio transformed microbial genes (alr-MG) had non-zero genomic correlations (rg) with at least one ADG-trait (ranging from |0.21| to |0.42|). Only a few alr-MG correlated with more than one ADG-trait, which suggests that a differential host-microbiome determinism underlies ADG at different stages. These alr-MG were involved in ribosomal biosynthesis, energy processes, sulphur and aminoacid metabolism and transport, or lipopolysaccharide signalling, among others. We selected two alternative subsets of 32 alr-MG that had a non-uniform or a uniform rg sign with all the ADG-traits, regardless of the rg magnitude, and used them to develop a microbiome-driven breeding strategy based on alr-MG only, or combined with ADG-traits, which was aimed at shaping the rumen microbiome towards increased ADG at all finishing stages. Combining alr-MG information with ADG records increased prediction accuracy of genomic estimated breeding values (GEBV) by 11 to 22% relative to the direct breeding strategy (using ADG-traits only), whereas using microbiome information, only, achieved lower accuracies (from 7 to 41%). Predicted selection responses varied consistently with accuracies. Restricting alr-MG based on their rg sign (uniform subset) did not yield a gain in the predicted response compared to the non-uniform subset, which is explained by the absence of alr-MG showing non-zero rg at least with more than one of the ADG-traits. CONCLUSIONS: Our work sheds light on the role of the microbial metabolism in the growth trajectory of beef cattle at the genomic level and provides insights into the potential benefits of using microbiome information in future genomic breeding programs to accurately estimate GEBV and increase ADG at each finishing stage in beef cattle.

KOunt: a reproducible KEGG orthologue abundance workflow.

SUMMARY: Accurate gene prediction is essential for successful metagenome analysis. We present KOunt, a Snakemake pipeline, that precisely quantifies KEGG orthologue abundance. AVAILABILITY AND IMPLEMENTATION: KOunt is available on GitHub: https://github.com/WatsonLab/KOunt. The KOunt reference database is available on figshare: https://doi.org/10.6084/m9.figshare.21269715. Test data are available at https://doi.org/10.6084/m9.figshare.22250152 and version 1.2.0 of KOunt at https://doi.org/10.6084/m9.figshare.23607834.

Microbiome-driven breeding strategy potentially improves beef fatty acid profile benefiting human health and reduces methane emissions.

BACKGROUND: Healthier ruminant products can be achieved by adequate manipulation of the rumen microbiota to increase the flux of beneficial fatty acids reaching host tissues. Genomic selection to modify the microbiome function provides a permanent and accumulative solution, which may have also favourable consequences in other traits of interest (e.g. methane emissions). Possibly due to a lack of data, this strategy has never been explored. RESULTS: This study provides a comprehensive identification of ruminal microbial mechanisms under host genomic influence that directly or indirectly affect the content of unsaturated fatty acids in beef associated with human dietary health benefits C18:3n-3, C20:5n-3, C22:5n-3, C22:6n-3 or cis-9, trans-11 C18:2 and trans-11 C18:1 in relation to hypercholesterolemic saturated fatty acids C12:0, C14:0 and C16:0, referred to as N3 and CLA indices. We first identified that ~27.6% (1002/3633) of the functional core additive log-ratio transformed microbial gene abundances (alr-MG) in the rumen were at least moderately host-genomically influenced (HGFC). Of these, 372 alr-MG were host-genomically correlated with the N3 index (n=290), CLA index (n=66) or with both (n=16), indicating that the HGFC influence on beef fatty acid composition is much more complex than the direct regulation of microbial lipolysis and biohydrogenation of dietary lipids and that N3 index variation is more strongly subjected to variations in the HGFC than CLA. Of these 372 alr-MG, 110 were correlated with the N3 and/or CLA index in the same direction, suggesting the opportunity for enhancement of both indices simultaneously through a microbiome-driven breeding strategy. These microbial genes were involved in microbial protein synthesis (aroF and serA), carbohydrate metabolism and transport (galT, msmX), lipopolysaccharide biosynthesis (kdsA, lpxD, lpxB), or flagellar synthesis (flgB, fliN) in certain genera within the Proteobacteria phyla (e.g. Serratia, Aeromonas). A microbiome-driven breeding strategy based on these microbial mechanisms as sole information criteria resulted in a positive selection response for both indices (1.36±0.24 and 0.79±0.21 sd of N3 and CLA indices, at 2.06 selection intensity). When evaluating the impact of our microbiome-driven breeding strategy to increase N3 and CLA indices on the environmental trait methane emissions (g/kg of dry matter intake), we obtained a correlated mitigation response of -0.41±0.12 sd. CONCLUSION: This research provides insight on the possibility of using the ruminal functional microbiome as information for host genomic selection, which could simultaneously improve several microbiome-driven traits of interest, in this study exemplified with meat quality traits and methane emissions. Video Abstract.

Bovine host genome acts on rumen microbiome function linked to methane emissions.

Our study provides substantial evidence that the host genome affects the comprehensive function of the microbiome in the rumen of bovines. Of 1,107/225/1,141 rumen microbial genera/metagenome assembled uncultured genomes (RUGs)/genes identified from whole metagenomics sequencing, 194/14/337 had significant host genomic effects (heritabilities ranging from 0.13 to 0.61), revealing that substantial variation of the microbiome is under host genomic control. We found 29/22/115 microbial genera/RUGs/genes host-genomically correlated (|0.59| to |0.93|) with emissions of the potent greenhouse gas methane (CH4), highlighting the strength of a common host genomic control of specific microbial processes and CH4. Only one of these microbial genes was directly involved in methanogenesis (cofG), whereas others were involved in providing substrates for archaea (e.g. bcd and pccB), important microbial interspecies communication mechanisms (ABC.PE.P), host-microbiome interaction (TSTA3) and genetic information processes (RP-L35). In our population, selection based on abundances of the 30 most informative microbial genes provided a mitigation potential of 17% of mean CH4 emissions per generation, which is higher than for selection based on measured CH4 using respiration chambers (13%), indicating the high potential of microbiome-driven breeding to cumulatively reduce CH4 emissions and mitigate climate change.

Potential of promotion of alleles by genome editing to improve quantitative traits in livestock breeding programs.

BACKGROUND: Genome editing (GE) is a method that enables specific nucleotides in the genome of an individual to be changed. To date, use of GE in livestock has focussed on simple traits that are controlled by a few quantitative trait nucleotides (QTN) with large effects. The aim of this study was to evaluate the potential of GE to improve quantitative traits that are controlled by many QTN, referred to here as promotion of alleles by genome editing (PAGE). METHODS: Multiple scenarios were simulated to test alternative PAGE strategies for a quantitative trait. They differed in (i) the number of edits per sire (0 to 100), (ii) the number of edits per generation (0 to 500), and (iii) the extent of use of PAGE (i.e. editing all sires or only a proportion of them). The base line scenario involved selecting individuals on true breeding values (i.e., genomic selection only (GS only)-genomic selection with perfect accuracy) for several generations. Alternative scenarios complemented this base line scenario with PAGE (GS + PAGE). The effect of different PAGE strategies was quantified by comparing response to selection, changes in allele frequencies, the number of distinct QTN edited, the sum of absolute effects of the edited QTN per generation, and inbreeding. RESULTS: Response to selection after 20 generations was between 1.08 and 4.12 times higher with GS + PAGE than with GS only. Increases in response to selection were larger with more edits per sire and more sires edited. When the total resources for PAGE were limited, editing a few sires for many QTN resulted in greater response to selection and inbreeding compared to editing many sires for a few QTN. Between the scenarios GS only and GS + PAGE, there was little difference in the average change in QTN allele frequencies, but there was a major difference for the QTN with the largest effects. The sum of the effects of the edited QTN decreased across generations. CONCLUSIONS: This study showed that PAGE has great potential for application in livestock breeding programs, but inbreeding needs to be managed.

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.

Applied animal genomics: results from the field.

Genomic selection (GS) is the use of statistical methods to estimate the genetic merit of a genotyped animal based on prediction equations derived from large ancestral populations with both phenotypes and genotypes. It has revolutionized the dairy cattle breeding industry and has been implemented with varying degrees of success in other animal breeding programs, including swine, poultry, and beef cattle. The findings of empirical field studies applying GS to the breeding sectors of these main animal protein industries are reviewed. Several translational considerations must be addressed before implementing GS in genetic improvement programs. These include determining and obtaining economically relevant phenotypes and determining the optimal size of the training population, cost-effective genotyping strategies, the practicality of field implementation, and the relative costs versus the benefits of the realized rate of genetic gain. GS may additionally change the optimal breeding scheme design, and studies that address this consideration are also reviewed briefly.

Genotype imputation to increase sample size in pedigreed populations.

Genotype imputation is a cost-effective way to increase the power of genomic selection or genome-wide association studies. While several genotype imputation algorithms are available, this chapter focuses on a heuristic algorithm, as implemented in the AlphaImpute software. This algorithm combines long-range phasing, haplotype library imputation, and segregation analysis and it is specifically designed to work with pedigreed populations.The chapter is organized in different sections. First the challenges related to genotype imputation in pedigreed populations are described, along with the specifics of the imputation algorithm used in AlphaImpute. In the second section, factors affecting the accuracy of genotype imputation using this algorithm are discussed. The different parameters that control AlphaImpute are detailed and examples of how to apply AlphaImpute are given.

Assessment of alternative genotyping strategies to maximize imputation accuracy at minimal cost.

BACKGROUND: Commercial breeding programs seek to maximise the rate of genetic gain while minimizing the costs of attaining that gain. Genomic information offers great potential to increase rates of genetic gain but it is expensive to generate. Low-cost genotyping strategies combined with genotype imputation offer dramatically reduced costs. However, both the costs and accuracy of imputation of these strategies are highly sensitive to several factors. The objective of this paper was to explore the cost and imputation accuracy of several alternative genotyping strategies in pedigreed populations. METHODS: Pedigree and genotype data from a commercial pig population were used. Several alternative genotyping strategies were explored. The strategies differed in the density of genotypes used for the ancestors and the individuals to be imputed. Parents, grandparents, and other relatives that were not descendants, were genotyped at high-density, low-density, or extremely low-density, and associated costs and imputation accuracies were evaluated. RESULTS: Imputation accuracy and cost were influenced by the alternative genotyping strategies. Given the mating ratios and the numbers of offspring produced by males and females, an optimized low-cost genotyping strategy for a commercial pig population could involve genotyping male parents at high-density, female parents at low-density (e.g. 3000 SNP), and selection candidates at very low-density (384 SNP). CONCLUSIONS: Among the selection candidates, 95.5% and 93.5% of the genotype variation contained in the high-density SNP panels were recovered using a genotyping strategy that costs respectively, $24.74 and $20.58 per candidate.

Selecting markers and evaluating coverage.

The availability of genetic markers in many species has enabled the analysis of marker-trait associations ranging from small genomic regions to genome-wide scale. An appropriate set of markers must be identified to meet the objectives of any research, using a custom discovery and selection approach or by using a commercial product. The key considerations in selecting markers are the quantity and the distribution across the genome. Though decisions about how many markers to use are often pragmatic, influenced by costs and available technology, an evaluation of the marker coverage is important in understanding how to design an effective genomic research study with reasonable expectations about the power to obtain desired results. An important parameter to evaluate coverage is linkage disequilibrium, which can be used to determine the appropriate number of markers for a particular analysis and is related to the proportion of variance that can be explained by a given marker, or power. Finally, the type of analysis used to identify marker-trait associations may depend on marker coverage as the optimal approach, from a statistical or computational standpoint, may differ with changes in marker number and distribution.

A common dataset for genomic analysis of livestock populations.

Although common datasets are an important resource for the scientific community and can be used to address important questions, genomic datasets of a meaningful size have not generally been available in livestock species. We describe a pig dataset that PIC (a Genus company) has made available for comparing genomic prediction methods. We also describe genomic evaluation of the data using methods that PIC considers best practice for predicting and validating genomic breeding values, and we discuss the impact of data structure on accuracy. The dataset contains 3534 individuals with high-density genotypes, phenotypes, and estimated breeding values for five traits. Genomic breeding values were calculated using BayesB, with phenotypes and de-regressed breeding values, and using a single-step genomic BLUP approach that combines information from genotyped and un-genotyped animals. The genomic breeding value accuracy increased with increased trait heritability and with increased relationship between training and validation. In nearly all cases, BayesB using de-regressed breeding values outperformed the other approaches, but the single-step evaluation performed only slightly worse. This dataset was useful for comparing methods for genomic prediction using real data. Our results indicate that validation approaches accounting for relatedness between populations can correct for potential overestimation of genomic breeding value accuracies, with implications for genotyping strategies to carry out genomic selection programs.

A phasing and imputation method for pedigreed populations that results in a single-stage genomic evaluation.

BACKGROUND: Efficient, robust, and accurate genotype imputation algorithms make large-scale application of genomic selection cost effective. An algorithm that imputes alleles or allele probabilities for all animals in the pedigree and for all genotyped single nucleotide polymorphisms (SNP) provides a framework to combine all pedigree, genomic, and phenotypic information into a single-stage genomic evaluation. METHODS: An algorithm was developed for imputation of genotypes in pedigreed populations that allows imputation for completely ungenotyped animals and for low-density genotyped animals, accommodates a wide variety of pedigree structures for genotyped animals, imputes unmapped SNP, and works for large datasets. The method involves simple phasing rules, long-range phasing and haplotype library imputation and segregation analysis. RESULTS: Imputation accuracy was high and computational cost was feasible for datasets with pedigrees of up to 25 000 animals. The resulting single-stage genomic evaluation increased the accuracy of estimated genomic breeding values compared to a scenario in which phenotypes on relatives that were not genotyped were ignored. CONCLUSIONS: The developed imputation algorithm and software and the resulting single-stage genomic evaluation method provide powerful new ways to exploit imputation and to obtain more accurate genetic evaluations.

Genotype imputation for the prediction of genomic breeding values in non-genotyped and low-density genotyped individuals.

BACKGROUND: There is wide interest in calculating genomic breeding values (GEBVs) in livestock using dense, genome-wide SNP data. The general framework for genomic selection assumes all individuals are genotyped at high-density, which may not be true in practice. Methods to add additional genotypes for individuals not genotyped at high density have the potential to increase GEBV accuracy with little or no additional cost. In this study a long haplotype library was created using a long range phasing algorithm and used in combination with segregation analysis to impute dense genotypes for non-genotyped dams in the training dataset (S1) and for non-genotyped or low-density genotyped individuals in the prediction dataset (S2), using the 14th QTL-MAS Workshop dataset. Alternative low-density scenarios were evaluated for accuracy of imputed genotypes and prediction of GEBVs. RESULTS: In S1, females in the training population were not genotyped and prediction individuals were either not genotyped or genotyped at low-density (evenly spaced at 2, 5 or 10 Mb). The proportion of correctly imputed genotypes for training females did not change when genotypes were added for individuals in the prediction set whereas the number of correctly imputed genotypes in the prediction set increased slightly (S1). The S2 scenario assumed the complete training set was genotyped for all SNPs and the prediction set was not genotyped or genotyped at low-density. The number of correctly imputed genotypes increased with genotyping density in the prediction set. Accuracy of genomic breeding values for the prediction set in each scenario were the correlation of GEBVs with true breeding values and were used to evaluate the potential loss in accuracy with reduced genotyping. For both S1 and S2 the GEBV accuracies were similar when the prediction set was not genotyped and increased with the addition of low-density genotypes, with the increase larger for S2 than S1. CONCLUSIONS: Genotype imputation using a long haplotype library and segregation analysis is promising for application in sparsely-genotyped pedigrees. The results of this study suggest that dense genotypes can be imputed for selection candidates with some loss in genomic breeding value accuracy, but with levels of accuracy higher than traditional BLUP estimated breeding values. Accurate genotype imputation would allow for a single low-density SNP panel to be used across traits.

Genomic breeding value prediction using three Bayesian methods and application to reduced density marker panels.

BACKGROUND: Bayesian approaches for predicting genomic breeding values (GEBV) have been proposed that allow for different variances for individual markers resulting in a shrinkage procedure that uses prior information to coerce negligible effects towards zero. These approaches have generally assumed application to high-density genotype data on all individuals, which may not be the case in practice. In this study, three approaches were compared for their predictive power in computing GEBV when training at high SNP marker density and predicting at high or low densities: the well- known Bayes-A, a generalization of Bayes-A where scale and degrees of freedom are estimated from the data (Student-t) and a Bayesian implementation of the Lasso method. Twelve scenarios were evaluated for predicting GEBV using low-density marker subsets, including selection of SNP based on genome spacing or size of additive effect and the inclusion of unknown genotype information in the form of genotype probabilities from pedigree and genotyped ancestors. RESULTS: The GEBV accuracy (calculated as correlation between GEBV and traditional breeding values) was highest for Lasso, followed by Student-t and then Bayes-A. When comparing GEBV to true breeding values, Student-t was most accurate, though differences were small. In general the shrinkage applied by the Lasso approach was less conservative than Bayes-A or Student-t, indicating that Lasso may be more sensitive to QTL with small effects. In the reduced-density marker subsets the ranking of the methods was generally consistent. Overall, low-density, evenly-spaced SNPs did a poor job of predicting GEBV, but SNPs selected based on additive effect size yielded accuracies similar to those at high density, even when coverage was low. The inclusion of genotype probabilities to the evenly-spaced subsets showed promising increases in accuracy and may be more useful in cases where many QTL of small effect are expected. CONCLUSIONS: In this dataset the Student-t approach slightly outperformed the other methods when predicting GEBV at both high and low density, but the Lasso method may have particular advantages in situations where many small QTL are expected. When markers were selected at low density based on genome spacing, the inclusion of genotype probabilities increased GEBV accuracy which would allow a single low- density marker panel to be used across traits.

Evaluation of a genome-wide approach to multiple marker association considering different marker densities.

BACKGROUND: Genome-wide approaches to analyze single nucleotide polymorphism (SNP) data have proliferated due to the increased availability and affordability of markers, but in practice a small number of markers may be selected from sets that do not approach dense genome-wide coverage. This study focused on a genome-wide approach to identify markers useful to a breeding program using a Bayesian method to estimate effects for markers distributed across the genome at varied densities. A simulated dataset containing 4665 individual phenotypes for a quantitative trait and genotypes for 6000 SNPs spaced in 0.1 cM increments across six chromosomes was analyzed using a Bayesian approach in which effects for all single markers are simultaneously estimated. The dataset was also analyzed with marker densities reduced to 0.5, 1.0, 2.0 and 5.0 cM. Type I errors were not a major concern but replications of each analysis were performed to determine acceptance of estimated marker effects. RESULTS: The Bayesian analysis of the original dataset was able to estimate genetic values for markers in a small number of regions while shrinking other marker effects to zero. Analysis of the reduced density datasets also showed clear signals in a small number of regions where some effects appeared to be distributed across multiple markers. Replicates of the analyses provided evidence for regions with moderate and large effects. CONCLUSION: A Bayesian multiple marker approach appears to be suitable for predicting genetic values, even with reduced density datasets where large numbers of markers are not yet available for many species. These predicted genetic values can be implemented in marker assisted selection programs.