Whole genome analysis of infectious bovine keratoconjunctivitis in Angus cattle using Bayesian threshold models

Alterations to the bovine bacterial ocular surface microbiome in the context of infectious bovine keratoconjunctivitis

BACKGROUND

Infectious bovine keratoconjunctivitis (IBK) is a common cause of morbidity in cattle, resulting in significant economic losses. This study aimed to characterize the bovine bacterial ocular surface microbiome (OSM) through conjunctival swab samples from Normal eyes and eyes with naturally acquired, active IBK across populations of cattle using a three-part approach, including bacterial culture, relative abundance (RA, 16 S rRNA gene sequencing), and semi-quantitative random forest modeling (real-time polymerase chain reaction (RT-PCR)).

RESULTS

Conjunctival swab samples were obtained from eyes individually classified as Normal (n = 376) or IBK (n = 228) based on clinical signs. Cattle unaffected by IBK and the unaffected eye in cattle with contralateral IBK were used to obtain Normal eye samples. Moraxella bovis was cultured from similar proportions of IBK (7/228, 3.07%) and Normal eyes (1/159, 0.63%) (p = 0.1481). Moraxella bovoculi was cultured more frequently (p < 0.0001) in IBK (59/228, 25.88%) than Normal (7/159, 4.40%) eyes. RA (via 16 S rRNA gene sequencing) of Actinobacteriota was significantly higher in Normal eyes (p = 0.0045). Corynebacterium variabile and Corynebacterium stationis (Actinobacteriota) were detected at significantly higher RA (p = 0.0008, p = 0.0025 respectively) in Normal eyes. Rothia nasimurium (Actinobacteriota) was detected at significantly higher RA in IBK eyes (p < 0.0001). Alpha-diversity index was not significantly different between IBK and Normal eyes (p > 0.05). Alpha-diversity indices for geographic location (p < 0.001), age (p < 0.0001), sex (p < 0.05) and breed (p < 0.01) and beta-diversity indices for geographic location (p < 0.001), disease status (p < 0.01), age (p < 0.001), sex (p < 0.001) and breed (p < 0.001) were significantly different between groups. Modeling of RT-PCR values reliably categorized the microbiome of IBK and Normal eyes; primers for Moraxella bovoculi, Moraxella bovis, and Staphylococcus spp. were consistently the most significant canonical variables in these models.

CONCLUSIONS

The results provide further evidence that multiple elements of the bovine bacterial OSM are altered in the context of IBK, indicating the involvement of a variety of bacteria in addition to Moraxella bovis, including Moraxella bovoculi and R. nasimurium, among others. Actinobacteriota RA is altered in IBK, providing possible opportunities for novel therapeutic interventions. While RT-PCR modeling provided limited further support for the involvement of Moraxella bovis in IBK, this was not overtly reflected in culture or RA results. Results also highlight the influence of geographic location and breed type (dairy or beef) on the bovine bacterial OSM. RT-PCR modeling reliably categorized samples as IBK or Normal.

Component Causes of Infectious Bovine Keratoconjunctivitis: The Role of Genetic Factors in the Epidemiology of Infectious Bovine Keratoconjunctivitis

The purpose of this article is to discuss the host as a cause of infectious bovine keratoconjunctivitis (IBK). The focus is on the host genetics rather than characteristics of the host, such as age, sex, and season of birth. From 4 conducted studies, estimates of IBK heritability are generally less than 0.15, except for some estimates for Herefords and Angus cattle around 0.2 and 1 study reporting a heritability of 0.33. These magnitudes of heritability are typically described as low to moderate. Quantitative trait locus on chromosome 1, 2, 12, 13, 20, and 21 has been associated with IBK resistance.

A review of traditional and machine learning methods applied to animal breeding

The current livestock management landscape is transitioning to a high-throughput digital era where large amounts of information captured by systems of electro-optical, acoustical, mechanical, and biosensors is stored and analyzed on a daily and hourly basis, and actionable decisions are made based on quantitative and qualitative analytic results. While traditional animal breeding prediction methods have been used with great success until recently, the deluge of information starts to create a computational and storage bottleneck that could lead to negative long-term impacts on herd management strategies if not handled properly. A plethora of machine learning approaches, successfully used in various industrial and scientific applications, made their way in the mainstream approaches for livestock breeding techniques, and current results show that such methods have the potential to match or surpass the traditional approaches, while most of the time they are more scalable from a computational and storage perspective. This article provides a succinct view on what traditional and novel prediction methods are currently used in the livestock breeding field, how successful they are, and how the future of the field looks in the new digital agriculture era.

A Benchmarking Between Deep Learning, Support Vector Machine and Bayesian Threshold Best Linear Unbiased Prediction for Predicting Ordinal Traits in Plant Breeding

Genomic selection is revolutionizing plant breeding. However, still lacking are better statistical models for ordinal phenotypes to improve the accuracy of the selection of candidate genotypes. For this reason, in this paper we explore the genomic based prediction performance of two popular machine learning methods: the Multi Layer Perceptron (MLP) and support vector machine (SVM) methods vs. the Bayesian threshold genomic best linear unbiased prediction (TGBLUP) model. We used the percentage of cases correctly classified (PCCC) as a metric to measure the prediction performance, and seven real data sets to evaluate the prediction accuracy, and found that the best predictions (in four out of the seven data sets) in terms of PCCC occurred under the TGLBUP model, while the worst occurred under the SVM method. Also, in general we found no statistical differences between using 1, 2 and 3 layers under the MLP models, which means that many times the conventional neuronal network model with only one layer is enough. However, although even that the TGBLUP model was better, we found that the predictions of MLP and SVM were very competitive with the advantage that the SVM was the most efficient in terms of the computational time required.

Genomic prediction of unordered categorical traits: an application to subpopulation assignment in German Warmblood horses

BACKGROUND

Categorical traits without ordinal representation of classes do not qualify for threshold models. Alternatively, the multinomial problem can be assessed by a sequence of independent binary contrasts using schemes such as one-vs-all or one-vs-one. Class probabilities can be arrived at by normalization or pair-wise coupling strategies. We assessed the predictive ability of whole-genome regression models and support vector machines for the classification of horses into four German Warmblood breeds.

RESULTS

Prediction accuracies of leave-one-out cross-validation were high and ranged from 0.75 to 0.97 depending on the binary classifier and breeds incorporated in the training. An analysis of the population structure using eigenvectors of the genomic relationship matrix revealed clustering of individuals beyond the given breed labels. Admixture between two breeds became apparent which had substantial impact on the prediction accuracies between those two breeds and also influenced the contrasts between other breeds.

CONCLUSIONS

Genomic prediction of unordered categorical traits was successfully applied to subpopulation assignment of German Warmblood horses. The applied methodology is a straightforward extension of existing binary threshold models for genomic prediction.

Threshold models for genome-enabled prediction of ordinal categorical traits in plant breeding

Categorical scores for disease susceptibility or resistance often are recorded in plant breeding. The aim of this study was to introduce genomic models for analyzing ordinal characters and to assess the predictive ability of genomic predictions for ordered categorical phenotypes using a threshold model counterpart of the Genomic Best Linear Unbiased Predictor (i.e., TGBLUP). The threshold model was used to relate a hypothetical underlying scale to the outward categorical response. We present an empirical application where a total of nine models, five without interaction and four with genomic × environment interaction (G×E) and genomic additive × additive × environment interaction (G×G×E), were used. We assessed the proposed models using data consisting of 278 maize lines genotyped with 46,347 single-nucleotide polymorphisms and evaluated for disease resistance [with ordinal scores from 1 (no disease) to 5 (complete infection)] in three environments (Colombia, Zimbabwe, and Mexico). Models with G×E captured a sizeable proportion of the total variability, which indicates the importance of introducing interaction to improve prediction accuracy. Relative to models based on main effects only, the models that included G×E achieved 9-14% gains in prediction accuracy; adding additive × additive interactions did not increase prediction accuracy consistently across locations.

Reduction in accuracy of genomic prediction for ordered categorical data compared to continuous observations

BACKGROUND

Accuracy of genomic prediction depends on number of records in the training population, heritability, effective population size, genetic architecture, and relatedness of training and validation populations. Many traits have ordered categories including reproductive performance and susceptibility or resistance to disease. Categorical scores are often recorded because they are easier to obtain than continuous observations. Bayesian linear regression has been extended to the threshold model for genomic prediction. The objective of this study was to quantify reductions in accuracy for ordinal categorical traits relative to continuous traits.

METHODS

Efficiency of genomic prediction was evaluated for heritabilities of 0.10, 0.25 or 0.50. Phenotypes were simulated for 2250 purebred animals using 50 QTL selected from actual 50k SNP (single nucleotide polymorphism) genotypes giving a proportion of causal to total loci of.0001. A Bayes C π threshold model simultaneously fitted all 50k markers except those that represented QTL. Estimated SNP effects were utilized to predict genomic breeding values in purebred (n = 239) or multibreed (n = 924) validation populations. Correlations between true and predicted genomic merit in validation populations were used to assess predictive ability.

RESULTS

Accuracies of genomic estimated breeding values ranged from 0.12 to 0.66 for purebred and from 0.04 to 0.53 for multibreed validation populations based on Bayes C π linear model analysis of the simulated underlying variable. Accuracies for ordinal categorical scores analyzed by the Bayes C π threshold model were 20% to 50% lower and ranged from 0.04 to 0.55 for purebred and from 0.01 to 0.44 for multibreed validation populations. Analysis of ordinal categorical scores using a linear model resulted in further reductions in accuracy.

CONCLUSIONS

Threshold traits result in markedly lower accuracy than a linear model on the underlying variable. To achieve an accuracy equal or greater than for continuous phenotypes with a training population of 1000 animals, a 2.25 fold increase in training population size was required for categorical scores fitted with the threshold model. The threshold model resulted in higher accuracies than the linear model and its advantage was greatest when training populations were smallest.

Reduction in accuracy of genomic prediction for ordered categorical data compared to continuous observations

Background

Accuracy of genomic prediction depends on number of records in the training population, heritability, effective population size, genetic architecture, and relatedness of training and validation populations. Many traits have ordered categories including reproductive performance and susceptibility or resistance to disease. Categorical scores are often recorded because they are easier to obtain than continuous observations. Bayesian linear regression has been extended to the threshold model for genomic prediction. The objective of this study was to quantify reductions in accuracy for ordinal categorical traits relative to continuous traits.

Methods

Efficiency of genomic prediction was evaluated for heritabilities of 0.10, 0.25 or 0.50. Phenotypes were simulated for 2250 purebred animals using 50 QTL selected from actual 50k SNP (single nucleotide polymorphism) genotypes giving a proportion of causal to total loci of.0001. A Bayes C π threshold model simultaneously fitted all 50k markers except those that represented QTL. Estimated SNP effects were utilized to predict genomic breeding values in purebred (n = 239) or multibreed (n = 924) validation populations. Correlations between true and predicted genomic merit in validation populations were used to assess predictive ability.

Results

Accuracies of genomic estimated breeding values ranged from 0.12 to 0.66 for purebred and from 0.04 to 0.53 for multibreed validation populations based on Bayes C π linear model analysis of the simulated underlying variable. Accuracies for ordinal categorical scores analyzed by the Bayes C π threshold model were 20% to 50% lower and ranged from 0.04 to 0.55 for purebred and from 0.01 to 0.44 for multibreed validation populations. Analysis of ordinal categorical scores using a linear model resulted in further reductions in accuracy.

Conclusions

Threshold traits result in markedly lower accuracy than a linear model on the underlying variable. To achieve an accuracy equal or greater than for continuous phenotypes with a training population of 1000 animals, a 2.25 fold increase in training population size was required for categorical scores fitted with the threshold model. The threshold model resulted in higher accuracies than the linear model and its advantage was greatest when training populations were smallest.

Recombination locations and rates in beef cattle assessed from parent-offspring pairs

BACKGROUND

Recombination events tend to occur in hotspots and vary in number among individuals. The presence of recombination influences the accuracy of haplotype phasing and the imputation of missing genotypes. Genes that influence genome-wide recombination rate have been discovered in mammals, yeast, and plants. Our aim was to investigate the influence of recombination on haplotype phasing, locate recombination hotspots, scan the genome for Quantitative Trait Loci (QTL) and identify candidate genes that influence recombination, and quantify the impact of recombination on the accuracy of genotype imputation in beef cattle.

METHODS

2775 Angus and 1485 Limousin parent-verified sire/offspring pairs were genotyped with the Illumina BovineSNP50 chip. Haplotype phasing was performed with DAGPHASE and BEAGLE using UMD3.1 assembly SNP (single nucleotide polymorphism) coordinates. Recombination events were detected by comparing the two reconstructed chromosomal haplotypes inherited by each offspring with those of their sires. Expected crossover probabilities were estimated assuming no interference and a binomial distribution for the frequency of crossovers. The BayesB approach for genome-wide association analysis implemented in the GenSel software was used to identify genomic regions harboring QTL with large effects on recombination. BEAGLE was used to impute Angus genotypes from a 7K subset to the 50K chip.

RESULTS

DAGPHASE was superior to BEAGLE in haplotype phasing, which indicates that linkage information from relatives can improve its accuracy. The estimated genetic length of the 29 bovine autosomes was 3097 cM, with a genome-wide recombination distance averaging 1.23 cM/Mb. 427 and 348 windows containing recombination hotspots were detected in Angus and Limousin, respectively, of which 166 were in common. Several significant SNPs and candidate genes, which influence genome-wide recombination were localized in QTL regions detected in the two breeds. High-recombination rates hinder the accuracy of haplotype phasing and genotype imputation.

CONCLUSIONS

Small population sizes, inadequate half-sib family sizes, recombination, gene conversion, genotyping errors, and map errors reduce the accuracy of haplotype phasing and genotype imputation. Candidate regions associated with recombination were identified in both breeds. Recombination analysis may improve the accuracy of haplotype phasing and genotype imputation from low- to high-density SNP panels.

Genome-wide association analysis and genomic prediction of Mycobacterium avium subspecies paratuberculosis infection in US Jersey cattle

Paratuberculosis (Johne’s disease), an enteric disorder in ruminants caused by Mycobacterium avium subspecies paratuberculosis (MAP), causes economic losses in excess of $200 million annually to the US dairy industry. To identify genomic regions underlying susceptibility to MAP infection in Jersey cattle, a case-control genome-wide association study (GWAS) was performed. Blood and fecal samples were collected from ∼5,000 mature cows in 30 commercial Jersey herds from across the US. Discovery data consisted of 450 cases and 439 controls genotyped with the Illumina BovineSNP50 BeadChip. Cases were animals with positive ELISA and fecal culture (FC) results. Controls were animals negative to both ELISA and FC tests that matched cases on birth date and herd. Validation data consisted of 180 animals including 90 cases (positive to FC) and 90 controls (negative to ELISA and FC), selected from discovery herds and genotyped by Illumina BovineLD BeadChip (∼7K SNPs). Two analytical approaches were used: single-marker GWAS using the GRAMMAR-GC method and Bayesian variable selection (Bayes C) using GenSel software. GRAMMAR-GC identified one SNP on BTA7 at 68 megabases (Mb) surpassing a significance threshold of 5×10⁻⁵. ARS-BFGL-NGS-11887 on BTA23 (27.7 Mb) accounted for the highest percentage of genetic variance (3.3%) in the Bayes C analysis. SNPs identified in common by GRAMMAR-GC and Bayes C in both discovery and combined data were mapped to BTA23 (27, 29 and 44 Mb), 3 (100, 101, 106 and 107 Mb) and 17 (57 Mb). Correspondence between results of GRAMMAR-GC and Bayes C was high (70–80% of most significant SNPs in common). These SNPs could potentially be associated with causal variants underlying susceptibility to MAP infection in Jersey cattle. Predictive performance of the model developed by Bayes C for prediction of infection status of animals in validation set was low (55% probability of correct ranking of paired case and control samples).

Overview of Statistical Methods for Genome-Wide Association Studies (GWAS)

This chapter provides an overview of statistical methods for genome-wide association studies (GWAS) in animals, plants, and humans. The simplest form of GWAS, a marker-by-marker analysis, is illustrated with a simple example. The problem of selecting a significance threshold that accounts for the large amount of multiple testing that occurs in GWAS is discussed. Population structure causes false positive associations in GWAS if not accounted for, and methods to deal with this are presented. Methodology for more complex models for GWAS, including haplotype-based approaches, accounting for identical by descent versus identical by state, and fitting all markers simultaneously are described and illustrated with examples.

Genome position specific priors for genomic prediction

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.

ggbio: an R package for extending the grammar of graphics for genomic data

We introduce ggbio, a new methodology to visualize and explore genomics annotations and high-throughput data. The plots provide detailed views of genomic regions, summary views of sequence alignments and splicing patterns, and genome-wide overviews with karyogram, circular and grand linear layouts. The methods leverage the statistical functionality available in R, the grammar of graphics and the data handling capabilities of the Bioconductor project. The plots are specified within a modular framework that enables users to construct plots in a systematic way, and are generated directly from Bioconductor data structures. The ggbio R package is available at http://www.bioconductor.org/packages/2.11/bioc/html/ggbio.html.

Scientific Opinion on the welfare of cattle kept for beef production and the welfare in intensive calf farming systems

Information given in previous Opinions "Welfare of cattle kept for beef production" (SCAHAW, 2001) and "The risks of poor welfare in intensive calf farming systems" (EFSA, 2006) is updated and recent scientific evidence on the topics reviewed. Risks of poor welfare are identified using a structured analysis, and issues not identified in the SCAHAW (2001) beef Opinion, especially effects of housing and management on enteric and respiratory diseases are reviewed. The Opinion covers all systems of beef production, although the welfare of suckler cows or breeding bulls is not considered. The Chapter on beef cattle presents new evidence and recommendations in relation to heat and cold stress, mutilations and pain management, digestive disorders linked to high concentrate feeds and respiratory disorders linked to overstocking, inadequate ventilation, mixing of animals and failure of early diagnosis and treatment. Major welfare problems in cattle kept for beef production, as identified by risk assessment, were respiratory diseases linked to overstocking, inadequate ventilation, mixing of animals and failure of early diagnosis and treatment, digestive disorders linked to intensive concentrate feeding, lack of physically effective fibre in the diet, and behavioural disorders linked to inadequate floor space, and co-mingling in the feedlot. Major hazards for white veal calves were considered to be iron-deficiency anaemia, a direct consequence of dietary iron restriction, enteric diseases linked to high intakes of liquid feed and inadequate intake of physically effective fibre, discomfort and behavioural disorders linked to inadequate floors and floor space.