Assessment of Poisson, logit, and linear models for genetic analysis of clinical mastitis in Norwegian Red cows

Insight into selective breeding for robustness based on field survival records: New genetic evaluation of survival traits in pacific white shrimp (Penaeus vannamei) breeding line

Survival can be considered a relatively ‘old’ trait in animal breeding, yet commonly neglected in aquaculture breeding because of the simple binary records and generally low heritability estimates. Developing routine genetic evaluation systems for survival traits however, will be important for breeding robust strains based on valuable field survival data. In the current study, linear multivariate animal model (LMA) was used for the genetic analysis of survival records from 2-year classes (BL2019 and BL2020) of pacific white shrimp (Penaeus vannamei) breeding lines with data collection of 52, 248 individuals from 481 fullsib families. During grow-out test period, 10 days intervals of survival data were considered as separate traits. Two survival definitions, binary survivability (S) and continuous survival in days (SL), were used for the genetic analysis of survival records to investigate; 1) whether adding more survival time information could improve estimation of genetic parameters; 2) the trajectory of survival heritability across time, and 3) patterns of genetic correlations of survival traits across time. Levels of heritability estimates for both S and SL were low (0.005–0.076), while heritability for survival day number was found to be similar with that of binary records at each observation time and were highly genetically correlated (r_g > 0.8). Heritability estimates of body weight (BW) for BL2019 and BL2020 were 0.486 and 0.373, respectively. Trajectories of survival heritability showed a gradual increase across the grow-out test period but slowed or reached a plateau during the later grow-out test period. Genetic correlations among survival traits in the grow-out tests were moderate to high, and the closer the times were between estimates, the higher were their genetic correlations. In contrast, genetic correlations between both survival traits and body weight were low but positive. Here we provide the first report on the trajectory of heritability estimates for survival traits across grow-out stage in aquaculture. Results will be useful for developing robust improved pacific white shrimp culture strains in selective breeding programs based on field survival data.

Genome-Wide Associations for Microscopic Differential Somatic Cell Count and Specific Mastitis Pathogens in Holstein Cows in Compost-Bedded Pack and Cubicle Farming Systems

Simple Summary

New free walk housing systems such as compost-bedded pack barns might positively influence animal welfare. However, udder health can be negatively affected due to the microbial environment in the pack. Udder health depends on many factors, such as the environment, the feed, the pathogen species, and the genetic mechanisms of the cow’s immune system. For a more precise evaluation of udder health, we examined novel traits including specific mastitis pathogens and differential somatic cell fractions in milk. In order to identify possible candidate genes for udder health, a genome-wide association study, including single-nucleotide polymorphisms (SNP) by housing system interactions (compost-bedded pack barn and conventional cubicle barn), was performed. We identified two potential candidate genes for the interaction effect in relation to udder health. The identified potential candidate gene HEMK1 (HemK methyltransferase family member 1) is involved in immune system development, and CHL1 (cell adhesion molecule L1 like) has an immunosuppressive effect during stress conditions. The results suggest housing system-specific breeding strategies in order to improve udder health in compost-bedded pack and conventional cubicle barns.

Abstract

The aim of the present study was to detect significant SNP (single-nucleotide polymorphism) effects and to annotate potential candidate genes for novel udder health traits in two different farming systems. We focused on specific mastitis pathogens and differential somatic cell fractions from 2198 udder quarters of 537 genotyped Holstein Friesian cows. The farming systems comprised compost-bedded pack and conventional cubicle barns. We developed a computer algorithm for genome-wide association studies allowing the estimation of main SNP effects plus consideration of SNPs by farming system interactions. With regard to the main effect, 35 significant SNPs were detected on 14 different chromosomes for the cell fractions and the pathogens. Six SNPs were significant for the interaction effect with the farming system for most of the udder health traits. We inferred two possible candidate genes based on significant SNP interactions. HEMK1 plays a role in the development of the immune system, depending on environmental stressors. CHL1 is regulated in relation to stress level and influences immune system mechanisms. The significant interactions indicate that gene activity can fluctuate depending on environmental stressors. Phenotypically, the prevalence of mastitis indicators differed between systems, with a notably lower prevalence of minor bacterial indicators in compost systems.

Genetic analysis of in-vitro embryo production traits in Dairy Gir cattle

The potential of dams as oocyte donors can be a selection criterion for animal breeding programs, but also an involuntary driver of the process. In both cases, it is important to determine genetic components influencing the outcome of in vitro embryo production (IVEP). The objective of the present study was to perform a detailed genetic analysis for in vitro embryo production traits in Dairy Gir cows. A dataset containing 11,450 records of ovum pick-up (OPU) and in vitro fertilization (IVF) procedures from 2684 Dairy Gir donors was evaluated. Analyzed traits were number (N_OV) and percentage (P_OV) of viable oocytes; number (N_GI) and percentage (P_GI) of grade I oocytes; number (N_EMB) and percentage (P_EMB) of viable embryos. All analyzes were performed using animal models by a Bayesian framework. Heritability estimates varied from 0.16 to 0.32 for count traits and from 0.01 to 0.06 for percentage traits. The proportion of the total variation represented by the additive genetic effect of sire (semen used in IVF) for N_EMB and P_EMB was 7% and 5% respectively. Associations between estimated breeding values from progeny tested bulls for IVEP traits, milk production, age at first calving and conformation traits were mainly low or close to zero. Results indicate that selection for IVEP traits is possible in Dairy Gir cattle and would not impair genetic progress for traits already considered as selection criteria. The N_OV seems to be a promising target trait. However, a selection index could help to avoid the use of sires with negative genetic merit for percentage traits, minimizing possible deterioration in the long term.

Survival of side grafts with scions from pure species Pinus engelmannii Carr. and the P. engelmannii × P. arizonica Engelm. var. arizonica hybrid

Grafting is one of the most widely used methods for vegetative propagation, particularly for multiplying trees considered important, but there has been little research done on the effect of hybridization on grafts from the genus Pinus. Sometimes hybrids show the ability to reproduce and adapt efficiently to the environment. However, they reduce the genetic gain of seed orchards. The objective in this research was to evaluate the effect of scion grafts from pure species donor Pinus engelmannii Carr. and from putative hybrid trees P. engelmannii × P. arizonica Engelm., grafted on rootstocks of pure species P. engelmannii, along with the effect of the position of the scion in the donor tree crown (upper third and middle third). The scions were collected from three trees of the pure species and three hybrids. In each tree, 20 scions were collected from each third of the crown evaluated. 120 side-veneer grafts were made at the beginning of spring (March) 2018. Variance analyses were performed to evaluate the treatments and adjustments of the Logit and Weibull models to obtain the probability of graft survival. Significant differences were found between the origins of scions (p < 0.0083, after Bonferroni correction), showing grafts with hybrid tree scions taking hold better. In addition, the probability of survival at 5 months after grafting with hybrid tree scions was greater (p < 0.0001) than in grafts with scions from trees of the pure species (Logit model), which coincides with the results of the Weibull model, which indicated that the probability of graft death with pure species donor tree scions is greater than for grafts with hybrid scions. There were no significant differences regarding the position of the scion in the donor tree crown.

Genetic correlation estimates between milk production traits, mastitis and different measures of somatic cells in Holstein cows

The objective of this study was to estimate genetic parameters including repeatability, heritability as well as genetic and environmental relationships between 305-day milk yield, milk fat and protein yield (Fat and Pro), milk fat and protein percentages (Fat% and Pro%), mastitis (Mast), number of mastitis occurrence and different measurements of somatic cell counts using linear and threshold animal as well as linear and threshold sire models in Holstein cows of Iran. Records of 33851 first lactation Holstein cows from five large dairy herds with calving dates from March 2002 to September 2014 were analysed, using Gibbs sampling methodology. Heritabilities of production traits estimated by linear animal model ranged from 0.14 (Fat%) to 0.29 (Pro%). Generally, udder health traits had low heritability (ranged from 0.005 to 0.10). Estimates of heritability for Mast using linear models were higher than those obtained with threshold models. However, in general estimates of heritabilities using threshold models were higher than those from linear models. There were unfavourable genetic correlations between production traits and Mast, which implies that breeding programs with emphasis on 305-day milk yield will experience deterioration in udder health. Despite low heritability of udder health traits, genetic variability exists for these traits that allow selecting superior animals and increasing resistance to Mast and animal welfare. Considering relatively high ratio of permanent environmental variance for Mast, culling decisions can be made with higher accuracy in order to reduce Mast incidence phenotypically over time.

Suitability of traits related to aggression and handleability for integration into pig breeding programmes: Genetic parameters and comparison between Gaussian and binary trait specifications

Changes in husbandry systems as well as consumers' increasing demands for animal welfare lead to increasing importance of traits such as handleability and aggressiveness in pigs. However, before using such novel traits for selection decisions, information on genetic parameters for these traits for the specific population is required. Therefore, weight gain and behaviour-related traits were recorded in 1004 pigs (814 Pietrain x German Landrace crossbred, 190 German Landrace purebred) at different ages. Behaviour indicators and tests were assessed and conducted, respectively under commercial farm conditions and included scoring of skin lesions (twice) and behaviour during backtests (twice), injections (once), handling (twice) and weighing (three times). Since behaviour scores often exhibit suboptimal statistical properties for parametric analyses, variance components were estimated using an animal model assuming a normal (Gaussian, GA; all traits) and additionally a binary distribution of variables (BI; using a logit-link function for all behaviour traits). Heritabilities for behavioural traits ranged from 0.02 ± 0.04 (finishing pig handling test; BI) to 0.36 ± 0.08 (backtest 2; GA) suggesting that some of the traits are potentially useful for genetic selection (e.g. finishing pig weighing test: h2 (GA) = 0.20 ± 0.07). Only minor differences were observed for results from binary and Gaussian analyses of the same traits suggesting that either approach might yield valid results. However, four-fold cross-validation using correlations between breeding values of a sub-set of animals for the sample trait finishing pig weighing score indicated slight superiority of the logit model (r = 0.85 ± 0.04 vs. r = 0.77 ± 0.03). Generally, only weak to moderate associations were found between behavioural reactions to the same test at different ages (rp ≤ 0.11 for weighing at different ages; rp = 0.30 but rg (GA) = 0.84 ± 0.11 for the backtests) as well as between reactions to different tests. Therefore, for inclusion of behaviour traits into breeding programmes, and considering high labour input required for some tests such as the backtest, it is recommended to assess behaviour during situations that are relevant and identical to practical conditions, while the use of indicator traits generally does not appear to be a very promising alternative.

Genome-Wide Association Study for Susceptibility to and Recoverability From Mastitis in Danish Holstein Cows

Because mastitis is very frequent and unavoidable, adding recovery information into the analysis for genetic evaluation of mastitis is of great interest from economical and animal welfare point of view. Here we have performed genome-wide association studies (GWAS) to identify associated single nucleotide polymorphisms (SNPs) and investigate the genetic background not only for susceptibility to – but also for recoverability from mastitis. Somatic cell count records from 993 Danish Holstein cows genotyped for a total of 39378 autosomal SNP markers were used for the association analysis. Single SNP regression analysis was performed using the statistical software package DMU. Substitution effect of each SNP was tested with a t-test and a genome-wide significance level of P-value < 10^-4 was used to declare significant SNP-trait association. A number of significant SNP variants were identified for both traits. Many of the SNP variants associated either with susceptibility to – or recoverability from mastitis were located in or very near to genes that have been reported for their role in the immune system. Genes involved in lymphocyte developments (e.g., MAST3 and STAB2) and genes involved in macrophage recruitment and regulation of inflammations (PDGFD and PTX3) were suggested as possible causal genes for susceptibility to – and recoverability from mastitis, respectively. However, this is the first GWAS study for recoverability from mastitis and our results need to be validated. The findings in the current study are, therefore, a starting point for further investigations in identifying causal genetic variants or chromosomal regions for both susceptibility to – and recoverability from mastitis.

Transgenerational epigenetics: Inheritance of global cytosine methylation and methylation-related epigenetic markers in the shrub Lavandula latifolia

PREMISE OF THE STUDY

The ecological and evolutionary significance of natural epigenetic variation (i.e., not based on DNA sequence variants) variation will depend critically on whether epigenetic states are transmitted from parents to offspring, but little is known on epigenetic inheritance in nonmodel plants.

METHODS

We present a quantitative analysis of transgenerational transmission of global DNA cytosine methylation (= proportion of all genomic cytosines that are methylated) and individual epigenetic markers (= methylation status of anonymous MSAP markers) in the shrub Lavandula latifolia. Methods based on parent-offspring correlations and parental variance component estimation were applied to epigenetic features of field-growing plants ('maternal parents') and greenhouse-grown progenies. Transmission of genetic markers (AFLP) was also assessed for reference.

KEY RESULTS

Maternal parents differed significantly in global DNA cytosine methylation (range = 21.7-36.7%). Greenhouse-grown maternal families differed significantly in global methylation, and their differences were significantly related to maternal origin. Methylation-sensitive amplified polymorphism (MSAP) markers exhibited significant transgenerational transmission, as denoted by significant maternal variance component of marker scores in greenhouse families and significant mother-offspring correlations of marker scores.

CONCLUSIONS

Although transmission-related measurements for global methylation and MSAP markers were quantitatively lower than those for AFLP markers taken as reference, this study has revealed extensive transgenerational transmission of genome-wide global cytosine methylation and anonymous epigenetic markers in L. latifolia. Similarity of results for global cytosine methylation and epigenetic markers lends robustness to this conclusion, and stresses the value of considering both types of information in epigenetic studies of nonmodel plants.

Genetic Parameters for Clinical Mastitis, Fertility and Somatic Cell Score in Czech Holstein Cattle

Cases of mastitis (CM) from 38,236 lactations belonging to 16,497 cows were recorded on seven farms in the Czech Republic from 1996 to 2014. Clinical mastitis was analyzed with linear animal model as an all-or-none trait for each recorded lactation (CM305) and separately for each trimester of lactation (CM1, CM2, and CM3). Bivariate linear animal models were used to estimate the genetic correlation between these CM traits and lactation means for somatic cell score (SCS305), the interval between calving and first insemination (INT) and days open (DO). Factors included in the linear model were parity, herd, year of calving, calving season, fixed linear and quadratic regression on age at first calving, fixed linear and quadratic regression on milk production in the corresponding parity, permanent environmental effect of the cow, and additive genetic effect of the cow. Estimated heritabilities of the CM traits ranged from 0.01 to 0.03. Permanent environmental effects accounted for approximately two-thirds of the phenotypic variance. Genetic correlations of SCS305 with CM traits were 0.85±0.029, 0.81±0.086, 0.82±0.087, and 0.67±0.088 for CM305, CM1, CM2, and CM3, respectively. Genetic correlations of INT with CM305, CM1, CM2, and CM3, respectively, were 0.22±0.065, 0.19±0.084, 0.20±0.121 and 0.15±0.121: and genetic correlations of DO and the four CM traits were 0.28±0.079, 0.26±0.101, 0.43±0.134, and 0.15±0.131. For the 140 sires in the dataset, Spearman rank correlations among breeding values for the four CM traits and for SCS305 were uniformly high at 0.99±0.001.

Modelling female fertility traits in beef cattle using linear and non-linear models

Female fertility traits are key components of the profitability of beef cattle production. However, these traits are difficult and expensive to measure, particularly under extensive pastoral conditions, and consequently, fertility records are in general scarce and somehow incomplete. Moreover, fertility traits are usually dominated by the effects of herd-year environment, and it is generally assumed that relatively small margins are kept for genetic improvement. New ways of modelling genetic variation in these traits are needed. Inspired in the methodological developments made by Prof. Daniel Gianola and co-workers, we assayed linear (Gaussian), Poisson, probit (threshold), censored Poisson and censored Gaussian models to three different kinds of endpoints, namely calving success (CS), number of days from first calving (CD) and number of failed oestrus (FE). For models involving FE and CS, non-linear models overperformed their linear counterparts. For models derived from CD, linear versions displayed better adjustment than the non-linear counterparts. Non-linear models showed consistently higher estimates of heritability and repeatability in all cases (h²  < 0.08 and r < 0.13, for linear models; h²  > 0.23 and r > 0.24, for non-linear models). While additive and permanent environment effects showed highly favourable correlations between all models (>0.789), consistency in selecting the 10% best sires showed important differences, mainly amongst the considered endpoints (FE, CS and CD). In consequence, endpoints should be considered as modelling different underlying genetic effects, with linear models more appropriate to describe CD and non-linear models better for FE and CS.

Bivariate threshold models for genetic evaluation of susceptibility to and ability to recover from mastitis in Danish Holstein cows

Mastitis in dairy cows is an unavoidable problem and genetic variation in recovery from mastitis, in addition to susceptibility, is therefore of interest. Genetic parameters for susceptibility to and recovery from mastitis were estimated for Danish Holstein-Friesian cows using data from automatic milking systems equipped with online somatic cell count measuring units. The somatic cell count measurements were converted to elevated mastitis risk, a continuous variable [on a (0-1) scale] indicating the risk of mastitis. Risk values >0.6 were assumed to indicate that a cow had mastitis. For each cow and lactation, the sequence of health states (mastitic or healthy) was converted to a weekly transition: 0 if the cow stayed within the same state and 1 if the cow changed state. The result was 2 series of transitions: one for healthy to diseased (HD, to model mastitis susceptibility) and the other for diseased to healthy (DH, to model recovery ability). The 2 series of transitions were analyzed with bivariate threshold models, including several systematic effects and a function of time. The model included effects of herd, parity, herd-test-week, permanent environment (to account for the repetitive nature of transition records from a cow) plus two time-varying effects (lactation stage and time within episode). In early lactation, there was an increased risk of getting mastitis but the risk remained stable afterwards. Mean recovery rate was 45% per lactation. Heritabilities were 0.07 [posterior mean of standard deviations (PSD) = 0.03] for HD and 0.08 (PSD = 0.03) for DH. The genetic correlation between HD and DH has a posterior mean of -0.83 (PSD = 0.13). Although susceptibility and recovery from mastitis are strongly negatively correlated, recovery can be considered as a new trait for selection.

Genetic heteroscedastic models for ordinal traits: application to sheep litter size

BACKGROUND

Classical genetic canalization models, which accommodate the mean and variance of a trait separately, provide a flexible approach to take heteroscedasticity for continuous traits into account. However, this model is not appropriate for discrete traits. The aim of this work was to propose heteroscedastic threshold models suitable for the genetic analysis of ordinal data.

METHODS

In order to first fit the mean and variance of ordinal traits separately, we extended the classical threshold model (TM) for discrete data by introducing non-genetic and genetic factors of heterogeneity on the variance of its underlying variable, which leads to a homothetic threshold model HTM and its alternative parameterization HTM' in which the thresholds of different individuals are linked by a homothetic-translation. Relaxing the constraint between the thresholds led us to propose an independent threshold model ITM that was more flexible than HTM' but required the estimation of more parameters. TM, HTM and ITM models were applied to study 19,671 records on litter size in Romane sheep.

RESULTS

Both HTM and ITM were able to disentangle the link between the mean and variance that holds in the classical homoscedastic threshold model. The results obtained for the litter size of Romane ewes showed that the data was best fitted with HTM compared to ITM and TM. The correlations between the observed and predicted variances were equal to 0.6 and 0.2 for HTM and TM, respectively. These analyses showed the existence of a genetic component for the heterogeneity of litter size in sheep that was taken into account in HTM.

CONCLUSIONS

HTM is the most suitable model to study the variability of litter size in sheep. It accommodates both the mean and variance separately while requiring the estimation of only a few parameters.

A Novel Statistical Model to Estimate Host Genetic Effects Affecting Disease Transmission

There is increasing recognition that genetic diversity can affect the spread of diseases, potentially affecting plant and livestock disease control as well as the emergence of human disease outbreaks. Nevertheless, even though computational tools can guide the control of infectious diseases, few epidemiological models can simultaneously accommodate the inherent individual heterogeneity in multiple infectious disease traits influencing disease transmission, such as the frequently modeled propensity to become infected and infectivity, which describes the host ability to transmit the infection to susceptible individuals. Furthermore, current quantitative genetic models fail to fully capture the heritable variation in host infectivity, mainly because they cannot accommodate the nonlinear infection dynamics underlying epidemiological data. We present in this article a novel statistical model and an inference method to estimate genetic parameters associated with both host susceptibility and infectivity. Our methodology combines quantitative genetic models of social interactions with stochastic processes to model the random, nonlinear, and dynamic nature of infections and uses adaptive Bayesian computational techniques to estimate the model parameters. Results using simulated epidemic data show that our model can accurately estimate heritabilities and genetic risks not only of susceptibility but also of infectivity, therefore exploring a trait whose heritable variation is currently ignored in disease genetics and can greatly influence the spread of infectious diseases. Our proposed methodology offers potential impacts in areas such as livestock disease control through selective breeding and also in predicting and controlling the emergence of disease outbreaks in human populations.

Predicting direct and indirect breeding values for survival time in laying hens using repeated measures

BACKGROUND

Minimizing bird losses is important in the commercial layer industry. Selection against mortality is challenging because heritability is low, censoring is high, and individual survival depends on social interactions among cage members. With cannibalism, mortality depends not only on an individual's own genes (direct genetic effects; DGE) but also on genes of its cage mates (indirect genetic effects; IGE). To date, studies using DGE-IGE models have focussed on survival time but their shortcomings are that censored records were considered as exact lengths of life and models assumed that IGE were continuously expressed by all cage members even after death. However, since dead animals no longer express IGE, IGE should ideally be time-dependent in the model. Neglecting censoring and timing of IGE expression may reduce accuracy of estimated breeding values (EBV). Thus, our aim was to improve prediction of breeding values for survival time in layers that present cannibalism.

METHODS

We considered four DGE-IGE models to predict survival time in layers. One model was an analysis of survival time and the three others treated survival in consecutive months as a repeated binomial trait (repeated measures models). We also tested whether EBV were improved by including timing of IGE expression in the analyses. Approximate EBV accuracies were calculated by cross-validation. The models were fitted to survival data on two purebred White Leghorn layer lines W1 and WB, each having monthly survival records over 13 months.

RESULTS

Including the timing of IGE expression in the DGE-IGE model reduced EBV accuracy compared to analysing survival time. EBV accuracy was higher when repeated measures models were used. However, there was no universal best model. Using repeated measures instead of analysing survival time increased EBV accuracy by 10 to 21 and 2 to 12 % for W1 and WB, respectively. We showed how EBV and variance components estimated with repeated measures models can be translated into survival time.

CONCLUSIONS

Our results suggest that prediction of breeding values for survival time in laying hens can be improved using repeated measures models. This is an important result since more accurate EBV contribute to higher rates of genetic gain.

A Bayesian generalized random regression model for estimating heritability using overdispersed count data

Background

Faecal egg counts are a common indicator of nematode infection and since it is a heritable trait, it provides a marker for selective breeding. However, since resistance to disease changes as the adaptive immune system develops, quantifying temporal changes in heritability could help improve selective breeding programs. Faecal egg counts can be extremely skewed and difficult to handle statistically. Therefore, previous heritability analyses have log transformed faecal egg counts to estimate heritability on a latent scale. However, such transformations may not always be appropriate. In addition, analyses of faecal egg counts have typically used univariate rather than multivariate analyses such as random regression that are appropriate when traits are correlated. We present a method for estimating the heritability of untransformed faecal egg counts over the grazing season using random regression.

Results

Replicating standard univariate analyses, we showed the dependence of heritability estimates on choice of transformation. Then, using a multitrait model, we exposed temporal correlations, highlighting the need for a random regression approach. Since random regression can sometimes involve the estimation of more parameters than observations or result in computationally intractable problems, we chose to investigate reduced rank random regression. Using standard software (WOMBAT), we discuss the estimation of variance components for log transformed data using both full and reduced rank analyses. Then, we modelled the untransformed data assuming it to be negative binomially distributed and used Metropolis Hastings to fit a generalized reduced rank random regression model with an additive genetic, permanent environmental and maternal effect. These three variance components explained more than 80 % of the total phenotypic variation, whereas the variance components for the log transformed data accounted for considerably less. The heritability, on a link scale, increased from around 0.25 at the beginning of the grazing season to around 0.4 at the end.

Conclusions

Random regressions are a useful tool for quantifying sources of variation across time. Our MCMC (Markov chain Monte Carlo) algorithm provides a flexible approach to fitting random regression models to non-normal data. Here we applied the algorithm to negative binomially distributed faecal egg count data, but this method is readily applicable to other types of overdispersed data.

Genetic analysis of clinical mastitis data for Holstein cattle in the Czech Republic

Abstract. Cases of mastitis were recorded from 22 812 lactations of 10 294 cows on seven farms in the Czech Republic from 2000 to 2012. The per cow number of clinical mastitis (CM) cases per lactation (CM1), number of days of CM per lactation (CM2), and CM considered as an all-or-none trait (CM3) with values of 0 (no CM case) or 1 (at least 1 CM case) were analyzed with linear animal models. Bivariate linear animal models were used for estimation of genetic correlations between CM traits and average lactation somatic cell score (SCS305), average 305-day milk (MY305), fat (FY305) and protein (PY305) yield, and interval between calving and first insemination (INT) and days open (DO). Factors included in the model of choice were parity, herd effect, year of calving, calving season, permanent environmental effect of the cow, and additive genetic effect of the cow. Estimated heritabilities for CM traits were in the range of 0.09 to 0.10. Genetic correlations of SCS305 with CM traits 1, 2, and 3 were 0.22 ± 0.062, 0.23 ± 0.064, and 0.29 ± 0.086, respectively; those of MY305 with the three CM traits were 0.80 ± 0.037, 0.79 ± 0.040, and 0.83 ± 0.038, respectively; those of INT with the three CM traits were 0.19 ± 0.087, 0.17 ± 0.089, and 0.26 ± 0.091, respectively; and those of DO with the three CM traits were 0.28 ± 0.089, 0.22 ± 0.091, and 0.27 ± 0.091, respectively. Knowledge of genetic parameters of mastitis incidence and assessment of the economic importance of the disease is necessary to design breeding programs to improve udder health.

Assessment of whole-genome regression for type II diabetes

Lifestyle and genetic factors play a large role in the development of Type 2 Diabetes (T2D). Despite the important role of genetic factors, genetic information is not incorporated into the clinical assessment of T2D risk. We assessed and compared Whole Genome Regression methods to predict the T2D status of 5,245 subjects from the Framingham Heart Study. For evaluating each method we constructed the following set of regression models: A clinical baseline model (CBM) which included non-genetic covariates only. CBM was extended by adding the first two marker-derived principal components and 65 SNPs identified by a recent GWAS consortium for T2D (M-65SNPs). Subsequently, it was further extended by adding 249,798 genome-wide SNPs from a high-density array. The Bayesian models used to incorporate genome-wide marker information as predictors were: Bayes A, Bayes Cπ, Bayesian LASSO (BL), and the Genomic Best Linear Unbiased Prediction (G-BLUP). Results included estimates of the genetic variance and heritability, genetic scores for T2D, and predictive ability evaluated in a 10-fold cross-validation. The predictive AUC estimates for CBM and M-65SNPs were: 0.668 and 0.684, respectively. We found evidence of contribution of genetic effects in T2D, as reflected in the genomic heritability estimates (0.492±0.066). The highest predictive AUC among the genome-wide marker Bayesian models was 0.681 for the Bayesian LASSO. Overall, the improvement in predictive ability was moderate and did not differ greatly among models that included genetic information. Approximately 58% of the total number of genetic variants was found to contribute to the overall genetic variation, indicating a complex genetic architecture for T2D. Our results suggest that the Bayes Cπ and the G-BLUP models with a large set of genome-wide markers could be used for predicting risk to T2D, as an alternative to using high-density arrays when selected markers from large consortiums for a given complex trait or disease are unavailable.

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.

Genetic parameters for feather pecking and aggressive behavior in a large F2-cross of laying hens using generalized linear mixed models

Feather pecking and aggressive pecking is a well-known problem in egg production. In the present study, genetic parameters for 4 feather-pecking-related traits were estimated using generalized linear mixed models. The traits were bouts of feather pecking delivered (FPD), bouts of feather pecking received (FPR), bouts of aggressive pecking delivered (APD), and bouts of aggressive pecking received (APR). An F2-design was established from 2 divergent selected founder lines. The lines were selected for low or high feather pecking for 10 generations. The number of F2 hens was 910. They were housed in pens with around 40 birds. Each pen was observed in 21 sessions of 20 min, distributed over 3 consecutive days. An animal model was applied that treated the bouts observed within 20 min as repeated observations. An over-dispersed Poisson distribution was assumed for observed counts and the link function was a log link. The model included a random animal effect, a random permanent environment effect, and a random day-by-hen effect. Residual variance was approximated on the link scale by the delta method. The results showed a heritability around 0.10 on the link scale for FPD and APD and of 0.04 for APR. The heritability of FPR was zero. For all behavior traits, substantial permanent environmental effects were observed. The approximate genetic correlation between FPD and APD (FPD and APR) was 0.81 (0.54). Egg production and feather eating records were collected on the same hens as well and were analyzed with a generalized linear mixed model, assuming a binomial distribution and using a probit link function. The heritability on the link scale for egg production was 0.40 and for feather eating 0.57. The approximate genetic correlation between FPD and egg production was 0.50 and between FPD and feather eating 0.73. Selection might help to reduce feather pecking, but this might result in an unfavorable correlated selection response reducing egg production. Feather eating and feather pecking are genetically correlated and this needs further investigation.

Animal models and integrated nested Laplace approximations

Animal models are generalized linear mixed models used in evolutionary biology and animal breeding to identify the genetic part of traits. Integrated Nested Laplace Approximation (INLA) is a methodology for making fast, nonsampling-based Bayesian inference for hierarchical Gaussian Markov models. In this article, we demonstrate that the INLA methodology can be used for many versions of Bayesian animal models. We analyze animal models for both synthetic case studies and house sparrow (Passer domesticus) population case studies with Gaussian, binomial, and Poisson likelihoods using INLA. Inference results are compared with results using Markov Chain Monte Carlo methods. For model choice we use difference in deviance information criteria (DIC). We suggest and show how to evaluate differences in DIC by comparing them with sampling results from simulation studies. We also introduce an R package, AnimalINLA, for easy and fast inference for Bayesian Animal models using INLA.

Preliminary genetic analyses of important musculoskeletal conditions of Thoroughbred racehorses in Hong Kong

A retrospective cohort study of important musculoskeletal conditions of Thoroughbred racehorses was conducted using health records generated over a 15 year period (n=5062, 1296 sires). The prevalence of each condition in the study population was: fracture, 13%; osteoarthritis, 10%; suspensory ligament injury, 10%; and tendon injury, 19%. Linear and logistic sire and animal regression models were built to describe the binary occurrence of these musculoskeletal conditions, and to evaluate the significance of possible environmental risk factors. The heritability of each condition was estimated using residual maximum likelihood (REML). Bivariate mixed models were used to generate estimates of genetic correlations between each pair of conditions. Heritability estimates of fracture, osteoarthritis, suspensory ligament and tendon injury were small to moderate (range: 0.01-0.20). Fracture was found to be positively genetically correlated with both osteoarthritis and suspensory ligament injury. These results suggest that there is a significant genetic component involved in the risk of the studied conditions. Due to positive genetic correlations, a reduction in prevalence of one of the correlated conditions may effect a reduction in risk of the other condition.

Genetic parameters for an alternative criterion to improve productive longevity of Nellore cows

Number of calvings at 53 mo (NC53) was proposed as an alternative selection criterion to improve productive longevity of Nellore cows. This study was carried out to estimate variance components for NC53 by assuming different models, so that the potential for using this selection criterion to improve fertility of Nellore cows could be assessed. Genetic correlations between NC53, number of calvings at 89 mo (NC89), and 2 selection indexes used in this breed were also estimated. The NC53 trait is moderately heritable (posterior mean heritability ≈ 0.17) and selecting for this criterion could improve productive longevity of Nellore cows. Greater response to selection is expected by fitting a threshold animal model for this trait, rather than a linear animal model. Greater accuracy of prediction for this criterion could be achieved by fitting a threshold-linear model, considering this trait and a selection index composed by traits evaluated at weaning and long-yearling.

Bias, accuracy, and impact of indirect genetic effects in infectious diseases

Selection for improved host response to infectious disease offers a desirable alternative to chemical treatment but has proven difficult in practice, due to low heritability estimates of disease traits. Disease data from field studies is often binary, indicating whether an individual has become infected or not following exposure to an infectious disease. Numerous studies have shown that from this data one can infer genetic variation in individuals’ underlying susceptibility. In a previous study, we showed that with an indirect genetic effect (IGE) model it is possible to capture some genetic variation in infectivity, if present, as well as in susceptibility. Infectivity is the propensity of transmitting infection upon contact with a susceptible individual. It is an important factor determining the severity of an epidemic. However, there are severe shortcomings with the Standard IGE models as they do not accommodate the dynamic nature of disease data. Here we adjust the Standard IGE model to (1) make expression of infectivity dependent on the individuals’ disease status (Case Model) and (2) to include timing of infection (Case-ordered Model). The models are evaluated by comparing impact of selection, bias, and accuracy of each model using simulated binary disease data. These were generated for populations with known variation in susceptibility and infectivity thus allowing comparisons between estimated and true breeding values. Overall the Case Model provided better estimates for host genetic susceptibility and infectivity compared to the Standard Model in terms of bias, impact, and accuracy. Furthermore, these estimates were strongly influenced by epidemiological characteristics. However, surprisingly, the Case-Ordered model performed considerably worse than the Standard and the Case Models, pointing toward limitations in incorporating disease dynamics into conventional variance component estimation methodology and software used in animal breeding.

Genetic relationships among mastitis and alternative somatic cell count traits in the first 3 lactations of Swedish Holsteins

The objectives of this study were to estimate heritabilities of, and genetic correlations among, clinical mastitis (CM), subclinical mastitis (SCM), and alternative somatic cell count (SCC) traits in the first 3 lactations of Swedish Holstein cows, and to estimate genetic correlations for the alternative traits across lactations. Data from cows having their first calving between 2002 and 2009 were used. The alternative SCC traits were based on information on CM and monthly test-day (TD) records of SCC traits of 178,613, 116,079, and 64,474 lactations in first, second, or third parity, respectively. Sires had an average of 230, 165, or 124 daughters in the data (parities 1, 2, or 3, respectively). Subclinical mastitis was defined as the number of periods with an SCC >150,000 cell/mL and without a treatment for CM. Average TD SCC between 5 and 150 d was used as a reference trait. The alternative SCC traits analyzed were 1) presence of at least 1 TD SCC between 41,000 and 80,000 cell/mL (TD41-80), 2) at least 1 TD SCC >500,000 cells/mL, 3) standard deviation of log SCC over the lactation, 4) number of infection peaks, and 5) average days diseased per peak. The same variables in different parities were treated as distinct traits. The statistical model considered the effects of herd-year, year, month, age at calving, animal, and residual. Heritability estimates were 0.07 to 0.08 for CM, 0.12 to 0.17 for SCM, and 0.14 for SCC150. For the alternative traits, heritability estimates were 0.12 to 0.17 for standard deviation of log SCC, TD SCC >500,000 cells/mL, and average days diseased per peak, and 0.06 to 0.10 for TD41-80 and number of infection peaks. Genetic correlations between CM with SCM were 0.62 to 0.74, and correlations for these traits with the alternative SCC traits were positive and very high (0.67 to 0.82 for CM, and 0.94 to 0.99 for SCM). Trait TD41-80 was the only alternative trait that showed negative, favorable, genetic correlations with CM (-0.22 to -0.50) and SCM (-0.48 to -0.85) because it is associated with healthy cows. Genetic correlations among the alternative traits in all 3 parities were high (0.93 to 0.99, 0.92 to 0.98, and 0.78 to 0.99, respectively). The only exception was TD41-80, which showed moderate to strong negative correlations with the rest of the traits. Genetic correlations of the same trait across parities were in general positive and very high (0.83 to 0.99). In conclusion, these alternative SCC traits could be used in practical breeding programs aiming to improve udder health in dairy cattle.

Indirect genetic effects and the spread of infectious disease: are we capturing the full heritable variation underlying disease prevalence?

Reducing disease prevalence through selection for host resistance offers a desirable alternative to chemical treatment. Selection for host resistance has proven difficult, however, due to low heritability estimates. These low estimates may be caused by a failure to capture all the relevant genetic variance in disease resistance, as genetic analysis currently is not taylored to estimate genetic variation in infectivity. Host infectivity is the propensity of transmitting infection upon contact with a susceptible individual, and can be regarded as an indirect effect to disease status. It may be caused by a combination of physiological and behavioural traits. Though genetic variation in infectivity is difficult to measure directly, Indirect Genetic Effect (IGE) models, also referred to as associative effects or social interaction models, allow the estimation of this variance from more readily available binary disease data (infected/non-infected). We therefore generated binary disease data from simulated populations with known amounts of variation in susceptibility and infectivity to test the adequacy of traditional and IGE models. Our results show that a conventional model fails to capture the genetic variation in infectivity inherent in populations with simulated infectivity. An IGE model, on the other hand, does capture some of the variation in infectivity. Comparison with expected genetic variance suggests that there is scope for further methodological improvement, and that potential responses to selection may be greater than values presented here. Nonetheless, selection using an index of estimated direct and indirect breeding values was shown to have a greater genetic selection differential and reduced future disease risk than traditional selection for resistance only. These findings suggest that if genetic variation in infectivity substantially contributes to disease transmission, then breeding designs which explicitly incorporate IGEs might help reduce disease prevalence.

Comparison of different models for genetic analysis of clinical mastitis in Austrian Fleckvieh dual-purpose cows

The performance of different models for genetic analyses of clinical mastitis in Austrian Fleckvieh dual-purpose cows was evaluated. The main objective was to compare threshold sire models (probit and logit) with linear sire and linear animal models using REML algorithm. For comparison, data were also analyzed using a Bayesian threshold sire model. The models were evaluated with respect to ranking of sires and their predictive ability in cross-validation. Only minor differences were observed in estimated variance components and heritability from Bayesian and REML probit models. Heritabilities for probit and logit models were 0.06 and 0.08, respectively, whereas heritabilities for linear sire and linear animal models were lower (0.02). Correlations among ranking of sires from threshold and linear sire models were high (>0.99), whereas correlations between any sire model (threshold or linear) and the linear animal model were slightly lower (0.96). The worst sires were ranked very similar across all models, whereas for the best sires some reranking occurred. Further, models were evaluated based on their ability to predict future data, which is one of the main concerns of animal breeders. The predictive ability of each model was determined by using 2 criteria: mean squared error and Pearson correlation between predicted and observed value. Overall, the 5 models did not differ in predictive ability. In contrast to expectations, sire models had the same predictive ability as animal models. Linear models were found to be robust toward departures from normality and performed equally well as threshold models.

Technical note: an R package for fitting generalized linear mixed models in animal breeding

Mixed models have been used extensively in quantitative genetics to study continuous and discrete traits. A standard quantitative genetic model proposes that the effects of levels of some random factor (e.g., sire) are correlated accordingly with their relationships. For this reason, routines for mixed models available in standard packages cannot be used for genetic analysis. The pedigreemm package of R was developed as an extension of the lme4 package, and allows mixed models with correlated random effects to be fitted for Gaussian, binary, and count responses. Following the method of Harville and Callanan (1989), a correlation between levels of the grouping factor (e.g., sire) is induced by post-multiplying the incidence matrix of the levels of this random factor by the Cholesky factor of the corresponding (co)variance matrix (e.g., the numerator relationship matrix between sires). Estimation methods available in pedigreemm include approximations to maximum likelihood and REML. This note describes the classes of models that can be fitted using pedigreemm and presents examples that illustrate its use.