Search

The Algorithm for Proven and Young ( APY) enables the implementation of single-step genomic BLUP (ss GBLUP) in large, genotyped populations by separating genotyped animals into core and non-core subsets and creating a computationally efficient inverse for the genomic relationship matrix ( G). As APY became the choice for large-scale genomic evaluations in BLUP-based methods, a common question is how to choose the animals in the core subset. We compared several core definitions to answer this question. Simulations comprised a moderately heritable trait for 95,010 animals and 50,000 genotypes for animals across five generations. Genotypes consisted of 25,500 SNP distributed across 15 chromosomes. Genotyping errors and missing pedigree were also mimicked. Core animals were defined based on individual generations, equal representation across generations, and at random. For a sufficiently large core size, core definitions had the same accuracies and biases, even if the core animals had imperfect genotypes. When genotyped animals had unknown parents, accuracy and bias were significantly better ( p ≤ .05) for random and across generation core definitions. [ABSTRACT FROM AUTHOR]

In genomic evaluations, it is desirable to have low computing cost while retaining high accuracy of evaluation for young animals. When the population is large but only few animals have phenotypes, especially for low heritability traits, the convergence rate of BLUP or single-step genomic BLUP (ssGBLUP) can be very slow. This study investigates the effect of pedigree truncation on convergence rate and solutions of ssGBLUP for data exhibiting slow convergence. The data consisted of 216,000, 221,000, 732,000, and 579,000 phenotypes on 4 traits. Heritabilities were less than 0.1 for 2 traits and greater than 0.2 for the other 2 traits. The full pedigree consisted of 2.4 million animals. Genotypes were available for 33,000 animals and consisted of 60,000 SNP. Two bivariate animal models were fit using pedigree-based BLUP or ssGBLUP. Either a regular or the algorithm for proven and young (APY) inverse was used for the genomic relationship matrix. Different pedigree depths were analyzed including full pedigree and 1 to 5 ancestral generations. Pedigree depths were defined as n ancestral generations for animals with phenotypes. The number of animals in the reduced pedigrees varied from 226,000 and 760,000 for 1 generation to 228,000 and 767,000 for 5 generations. Genomic EBV (GEBV) for genotyped animals had correlations greater than 0.99 between runs with the full and reduced pedigrees with 2 to 5 generations. A single generation of pedigree was not sufficient to obtain the same GEBV as full pedigree. The convergence rate was the worst with the full pedigree and generally improved with reduced pedigrees. Using ssGBLUP with the APY inverse improved convergence without affecting accuracy. Reducing pedigrees and the APY are important tools to reduce the computational cost in the implementation of ssGBLUP. [ABSTRACT FROM AUTHOR]

The purpose of this study is review and evaluation of computing methods used in genomic selection for animal breeding. Commonly used models include SNP BLUP with extensions (BayesA, etc), genomic BLUP (GBLUP) and single-step GBLUP (ssGBLUP). These models are applied for genomewide association studies (GWAS), genomic prediction and parameter estimation. Solving methods include finite Cholesky decomposition possibly with a sparse implementation, and iterative Gauss–Seidel (GS) or preconditioned conjugate gradient (PCG), the last two methods possibly with iteration on data. Details are provided that can drastically decrease some computations. For SNP BLUP especially with sampling and large number of SNP, the only choice is GS with iteration on data and adjustment of residuals. If only solutions are required, PCG by iteration on data is a clear choice. A genomic relationship matrix (GRM) has limited dimensionality due to small effective population size, resulting in infinite number of generalized inverses of GRM for large genotyped populations. A specific inverse called APY requires only a small fraction of GRM, is sparse and can be computed and stored at a low cost for millions of animals. With APY inverse and PCG iteration, GBLUP and ssGBLUP can be applied to any population. Both tools can be applied to GWAS. When the system of equations is sparse but contains dense blocks, a recently developed package for sparse Cholesky decomposition and sparse inversion called YAMS has greatly improved performance over packages where such blocks were treated as sparse. With YAMS, GREML and possibly single-step GREML can be applied to populations with >50 000 genotyped animals. From a computational perspective, genomic selection is becoming a mature methodology. [ABSTRACT FROM PUBLISHER]

This paper evaluates an efficient implementation to multiply the inverse of a numerator relationship matrix for genotyped animals (A22-1) by a vector (q). The computation is required for solving mixed model equations in single-step genomic BLUP (ssGBLUP) with the preconditioned conjugate gradient (PCG). The inverse can be decomposed into sparse matrices that are blocks of the sparse inverse of a numerator relationship matrix (A-1) including genotyped animals and their ancestors. The elements of A-1 were rapidly calculated with the Henderson's rule and stored as sparse matrices in memory. Implementation of A22-1 q was by a series of sparse matrix–vector multiplications. Diagonal elements of A22-1, which were required as preconditioners in PCG, were approximated with a Monte Carlo method using 1,000 samples. The efficient implementation of A22-1 q was compared with explicit inversion of A22 with 3 data sets including about 15,000, 81,000, and 570,000 genotyped animals selected from populations with 213,000, 8.2 million, and 10.7 million pedigree animals, respectively. The explicit inversion required 1.8 GB, 49 GB, and 2,415 GB (estimated) of memory, respectively, and 42 s, 56 min, and 13.5 d (estimated), respectively, for the computations. The efficient implementation required <1 MB, 2.9 GB, and 2.3 GB of memory, respectively, and <1 sec, 3 min, and 5 min, respectively, for setting up. Only <1 sec was required for the multiplication in each PCG iteration for any data sets. When the equations in ssGBLUP are solved with the PCG algorithm, A22-1 is no longer a limiting factor in the computations. [ABSTRACT FROM AUTHOR]

The purposes of this study were to analyze the impact of seasonal losses due to heat stress in pigs from different breeds raised in different environments and to evaluate the accuracy improvement from adding genomic information to genetic evaluations. Data were available for 2 different swine populations: purebred Duroc animals raised in Texas and North Carolina and commercial crosses of Duroc and F1 females (Landrace × Large White) raised in Missouri and North Carolina; pedigrees provided links for animals from different states. Pedigree information was available for 553,442 animals, of which 8,232 pure breeds were genotyped. Traits were BW at 170 d for purebred animals and HCW for crossbred animals. Analyses were done with an animal model as either single- or 2-trait models using phenotypes measured in different states as separate traits. Additionally, reaction norm models were fitted for 1 or 2 traits using heat load index as a covariable. Heat load was calculated as temperature-humidity index greater than 70 and was averaged over 30 d prior to data collection. Variance components were estimated with average information REML, and EBV and genomic EBV (GEBV) with BLUP or single-step genomic BLUP (ssGBLUP). Validation was assessed for 146 genotyped sires with progeny in the last generation. Accuracy was calculated as a correlation between EBV and GEBV using reduced data (all animals, except the last generation) and using complete data. Heritability estimates for purebred animals were similar across states (varying from 0.23 to 0.26), and reaction norm models did not show evidence of a heat stress effect. Genetic correlations between states for heat loads were always strong (>0.91). For crossbred animals, no differences in heritability were found in single- or 2-trait analysis (from 0.17 to 0.18), and genetic correlations between states were moderate (0.43). In the reaction norm for crossbreeds, heritabilities ranged from 0.15 to 0.30 and genetic correlations between heat loads were as weak as 0.36, with heat load ranging from 0 to 12. Accuracies with ssGBLUP were, on average, 25% greater than with BLUP. Accuracies were greater in 2-trait reaction norm models and at extreme heat load values. Impacts of seasonality are evident only for crossbred animals. Genomic information can help producers mitigate heat stress in swine by identifying superior sires that are more resistant to heat stress. [ABSTRACT FROM AUTHOR]

Animal feeding is the most important economic component of beef production systems. Selection for feed efficiency has not been effective mainly due to difficult and high costs to obtain the phenotypes. The application of genomic selection using SNP can decrease the cost of animal evaluation as well as the generation interval. The objective of this study was to compare methods for genomic evaluation of feed efficiency traits using different cross-validation layouts in an experimental beef cattle population genotyped for a high-density SNP panel (BovineHD BeadChip assay 700k, Illumina Inc., San Diego, CA). After quality control, a total of 437,197 SNP genotypes were available for 761 Nelore animals from the Institute of Animal Science, Sertãozinho, São Paulo, Brazil. The studied traits were residual feed intake, feed conversion ratio, ADG, and DMI. Methods of analysis were traditional BLUP, single-step genomic BLUP (ssGBLUP), genomic BLUP (GBLUP), and a Bayesian regression method (BayesCπ). Direct genomic values (DGV) from the last 2 methods were compared directly or in an index that combines DGV with parent average. Three cross-validation approaches were used to validate the models: 1) YOUNG, in which the partition into training and testing sets was based on year of birth and testing animals were born after 2010; 2) UNREL, in which the data set was split into 3 less related subsets and the validation was done in each subset a time; and 3) RANDOM, in which the data set was randomly divided into 4 subsets (considering the contemporary groups) and the validation was done in each subset at a time. On average, the RANDOM design provided the most accurate predictions. Average accuracies ranged from 0.10 to 0.58 using BLUP, from 0.09 to 0.48 using GBLUP, from 0.06 to 0.49 using BayesCπ, and from 0.22 to 0.49 using ssGBLUP. The most accurate and consistent predictions were obtained using ssGBLUP for all analyzed traits. The ssGBLUP seems to be more suitable to obtain genomic predictions for feed efficiency traits on an experimental population of genotyped animals. [ABSTRACT FROM AUTHOR]

The purpose of this study was to examine accuracy of genomic selection via single-step genomic BLUP (ss GBLUP) when the direct inverse of the genomic relationship matrix ( G) is replaced by an approximation of G−1 based on recursions for young genotyped animals conditioned on a subset of proven animals, termed algorithm for proven and young animals ( APY). With the efficient implementation, this algorithm has a cubic cost with proven animals and linear with young animals. Ten duplicate data sets mimicking a dairy cattle population were simulated. In a first scenario, genomic information for 20k genotyped bulls, divided in 7k proven and 13k young bulls, was generated for each replicate. In a second scenario, 5k genotyped cows with phenotypes were included in the analysis as young animals. Accuracies (average for the 10 replicates) in regular EBV were 0.72 and 0.34 for proven and young animals, respectively. When genomic information was included, they increased to 0.75 and 0.50. No differences between genomic EBV ( GEBV) obtained with the regular G−1 and the approximated G−1 via the recursive method were observed. In the second scenario, accuracies in GEBV (0.76, 0.51 and 0.59 for proven bulls, young males and young females, respectively) were also higher than those in EBV (0.72, 0.35 and 0.49). Again, no differences between GEBV with regular G−1 and with recursions were observed. With the recursive algorithm, the number of iterations to achieve convergence was reduced from 227 to 206 in the first scenario and from 232 to 209 in the second scenario. Cows can be treated as young animals in APY without reducing the accuracy. The proposed algorithm can be implemented to reduce computing costs and to overcome current limitations on the number of genotyped animals in the ss GBLUP method. [ABSTRACT FROM AUTHOR]

Predictive ability of genomic EBV when using single-step genomic BLUP (ssGBLUP) in Angus cattle was investigated. Over 6 million records were available on birth weight (BiW) and weaning weight (WW), almost 3.4 million on postweaning gain (PWG), and over 1.3 million on calving ease (CE). Genomic information was available on, at most, 51,883 animals, which included high and low EBV accuracy animals. Traditional EBV was computed by BLUP and genomic EBV by ssGBLUP and indirect prediction based on SNP effects was derived from ssGBLUP; SNP effects were calculated based on the following reference populations: ref_2k (contains top bulls and top cows that had an EBV accuracy for BiW ≥0.85), ref_8k (contains all parents that were genotyped), and ref_33k (contains all genotyped animals born up to 2012). Indirect prediction was obtained as direct genomic value (DGV) or as an index of DGV and parent average (PA). Additionally, runs with ssGBLUP used the inverse of the genomic relationship matrix calculated by an algorithm for proven and young animals (APY) that uses recursions on a small subset of reference animals. An extra reference subset included 3,872 genotyped parents of genotyped animals (ref_4k). Cross-validation was used to assess predictive ability on a validation population of 18,721 animals born in 2013. Computations for growth traits used multiple-trait linear model and, for CE, a bivariate CE-BiW threshold-linear model. With BLUP, predictivities were 0.29, 0.34, 0.23, and 0.12 for BiW, WW, PWG, and CE, respectively. With ssGBLUP and ref_2k, predictivities were 0.34, 0.35, 0.27, and 0.13 for BiW, WW, PWG, and CE, respectively, and with ssGBLUP and ref_33k, predictivities were 0.39, 0.38, 0.29, and 0.13 for BiW, WW, PWG, and CE, respectively. Low predictivity for CE was due to low incidence rate of difficult calving. Indirect predictions with ref_33k were as accurate as with full ssGBLUP. Using the APY and recursions on ref_4k gave 88% gains of full ssGBLUP and using the APY and recursions on ref_8k gave 97% gains of full ssGBLUP. Genomic evaluation in beef cattle with ssGBLUP is feasible while keeping the models (maternal, multiple trait, and threshold) already used in regular BLUP. Gains in predictivity are dependent on the composition of the reference population. Indirect predictions via SNP effects derived from ssGBLUP allow for accurate genomic predictions on young animals, with no advantage of including PA in the index if the reference population is large. With the APY conditioning on about 10,000 reference animals, ssGBLUP is potentially applicable to a large number of genotyped animals without compromising predictive ability. [ABSTRACT FROM AUTHOR]

Computing the inverse of the genomic relationship matrix using recursion was investigated. A traditional algorithm to invert the numerator relationship matrix is based on the observation that the conditional expectation for an additive effect of 1 animal given the effects of all other animals depends on the effects of its sire and dam only, each with a coefficient of 0.5. With genomic relationships, such an expectation depends on all other genotyped animals, and the coefficients do not have any set value. For each animal, the coefficients plus the conditional variance can be called a genomic recursion. If such recursions are known, the mixed model equations can be solved without explicitly creating the inverse of the genomic relationship matrix. Several algorithms were developed to create genomic recursions. In an algorithm with sequential updates, genomic recursions are created animal by animal. That algorithm can also be used to update a known inverse of a genomic relationship matrix for additional genotypes. In an algorithm with forward updates, a newly computed recursion is immediately applied to update recursions for remaining animals. The computing costs for both algorithms depend on the sparsity pattern of the genomic recursions, but are lower or equal than for regular inversion. An algorithm for proven and young animals assumes that the genomic recursions for young animals contain coefficients only for proven animals. Such an algorithm generates exact genomic EBV in genomic BLUP and is an approximation in single-step genomic BLUP. That algorithm has a cubic cost for the number of proven animals and a linear cost for the number of young animals. The genomic recursions can provide new insight into genomic evaluation and possibly reduce costs of genetic predictions with extremely large numbers of genotypes. [ABSTRACT FROM AUTHOR]

Data sets of US Holsteins, Israeli Holsteins, and pigs from PIC (a Genus company, Hendersonville, TN) were used to evaluate the effect of different numbers of generations on ability to predict genomic breeding values of young genotyped animals. The influence of including only 2 generations of ancestors (A2) or all ancestors (Af) was also investigated. A total of 34,506 US Holsteins, 1,305 Israeli Holsteins, and 5,236 pigs were genotyped. The evaluations were computed by traditional BLUP and single-step genomic BLUP, and computing performance was assessed for the latter method. For the 2 Holstein data sets, coefficients of determination (R2) and regression (δ) of deregressed evaluations from a full data set with records up to 2011 on estimated breeding values and genomic estimated breeding values from the truncated data sets were computed. The thresholds for data deletion were set by intervals of 5 yr, based on the average generation interval in dairy cattle. For the PIC data set, correlations between corrected phenotypes and estimated or genomic estimated breeding values were used to evaluate predictive ability on young animals born in 2010 and 2011. The reduced data set contained data up to 2009, and the thresholds were set based on an average generation interval of 3 yr. The number of generations that could be deleted without a reduction in accuracy depended on data structure and trait. For US Holsteins, removing 3 and 4 generations of data did not reduce accuracy of evaluations for final score in Af and A2 scenarios, respectively. For Israeli Holsteins, the accuracies for milk, fat, and protein yields were the highest when only phenotypes recorded in 2000 and later were included and full pedigrees were applied. Of the 135 Israeli bulls with genotypes (validation set) and daughter records only in the complete data set, 38 and 97 were sons of Israeli and foreign bulls, respectively. Although more phenotypic data increased the prediction accuracy for sons of Israeli bulls, the reverse was true for sons of foreign bulls. Also, more phenotypic data caused large inflation of genomic estimated breeding values for sons of foreign bulls, whereas the opposite was true with the deletion of all but the most recent phenotypic data. Results for protein and fat percentage were different from those for milk, fat, and protein yields; however, relatively, the changes in coefficients of determination and regression were smaller for percentage traits. For PIC data set, removing data from up to 5 generations did not erode predictive ability for genotyped animals for the 2 reproductive traits used in validation. Given the data used in this study, truncating old data reduces computation requirements but does not decrease the accuracy. For small populations that include local and imported animals, truncation may be beneficial for one group of animals and detrimental to another group. [ABSTRACT FROM AUTHOR]

Methods for genomic prediction were evaluated for an Israeli Holstein dairy population of 713,686 cows and 1,305 progeny-tested bulls with genotypes. Inclusion of genotypes of 343 elite cows in an evaluation method that considers pedigree, phenotypes, and genotypes simultaneously was also evaluated. Two data sets were available: a complete data set with production records from 1985 through 2011, and a reduced data set with records after 2006 deleted. For each production trait, a multitrait animal model was used to compute traditional genetic evaluations for parities 1 through 3 as separate traits. Evaluations were calculated for the reduced and complete data sets. The evaluations from the reduced data set were used to calculate parent average for validation bulls, which was the benchmark for comparing gain in predictive ability from genomics. Genomic predictions for bulls in 2006 were calculated using a Bayesian regression method (BayesC), genomic BLUP (GBLUP), single-step GBLUP (ssGBLUP), and weighted ssGBLUP (WssGBLUP). Predictions using BayesC and GBLUP were calculated either with or without an index that included parent average. Genomic predictions that included elite cow genotypes were calculated using ssGBLUP and WssGBLUP. Predictive ability was assessed by coefficients of determination (R²) and regressions of predictions of 135 validation bulls with no daughters in 2006 on deregressed evaluations of those bulls in 2011. A reduction in R² and regression coefficients was observed from parities 1 through 3. Fat and protein yields had the lowest R for all the methods. On average, R² was lowest for parent averages, followed by GBLUP, BayesC, ssGBLUP, and WssGBLUP. For some traits, R² for direct genomic values from BayesC and GBLUP were lower than those for parent averages. Genomic estimated breeding values using ssGBLUP were the least biased, and this method appears to be a suitable tool for genomic evaluation of a small genotyped population, as it automatically accounts for parental index, allows for inclusion of female genomic information without preadjustments in evaluations, and uses the same model as in traditional evaluations. Weighted ssGBLUP has the potential for higher evaluation accuracy. [ABSTRACT FROM AUTHOR]

Currently, genomic evaluations use multiple-step procedures, which are prone to biases and errors. A single-step procedure may be applicable when genomic predictions can be obtained by modifying the numerator relationship matrix A to H = A + AΔ, where AΔ includes deviations from expected relationships. However, the traditional mixed model equations require H-1, which is usually difficult to obtain for large pedigrees. The computations with H are feasible when the mixed model equations are expressed in an alternate form that also applies for singular H and when those equations are solved by the conjugate gradient techniques. Then the only computations involving H are in the form of Aq or AΔq, where q is a vector. The alternative equations have a nonsymmetric left-hand side. Computing AΔq is inexpensive when the number of nonzeros in AΔ is small, and the product Aq can be calculated efficiently in linear time using an indirect algorithm. Generalizations to more complicated models are proposed. The data included 10.2 million final scores on 6.2 million Holsteins and were analyzed by a repeatability model. Comparisons involved the regular and the alternative equations. The model for the second case included simulated A?. Solutions were obtained by the preconditioned conjugate gradient algorithm, which works only with symmetric matrices, and by the bi-conjugate gradient stabilized algorithm, which also works with nonsymmetric matrices. The convergence rate associated with the nonsymmetric solvers was slightly better than that with the symmetric solver for the original equations, although the time per round was twice as much for the nonsymmetric solvers. The convergence rate associated with the alternative equations ranged from 2 times lower without AΔ to 3 times lower for the largest simulated AΔ. When the information attributable to genomics can be expressed as modifications to the numerator relationship matrix, the proposed methodology may allow the upgrading of an existing evaluation to incorporate the genomic information. [ABSTRACT FROM AUTHOR]

Genome-wide genetic evaluation might involve the computation of BLUP-like estimations, potentially including thousands of covariates (i.e., single-nucleotide polymorphism markers) for each record. This implies dense Henderson's mixed-model equations and considerable computing resources in time and storage, even for a few thousand records. Possible computing options include the type of storage and the solving algorithm. This work evaluated several computing options, including half-stored Cholesky decomposition, Gauss-Seidel, and 3 matrix-free strategies: Gauss-Seidel, Gauss-Seidel with residuals update, and preconditioned conjugate gradients. Matrix-free Gauss-Seidel with residuals update adjusts the residuals after computing the solution for each effect. This avoids adjusting the left-hand side of the equations by all other effects at every step of the algorithm and saves considerable computing time. Any Gauss-Seidel algorithm can easily be extended for variance component estimation by Markov chain- Monte Carlo. Let m and n be the number of records and markers, respectively. Computing time for Cholesky decomposition is proportional to n³. Computing times per round are proportional to mn² in matrix-free Gauss- Seidel, to n² for half-stored Gauss-Seidel, and to n and m for the rest of the algorithms. Algorithms were tested on a real mouse data set, which included 1,928 records and 10,946 single-nucleotide polymorphism markers. Computing times were in the order of a few minutes for Gauss-Seidel with residuals update and preconditioned conjugate gradients, more than 1 h for half-stored Gauss-Seidel, 2 h for Cholesky decomposition, and 4 d for matrix-free Gauss-Seidel. Preconditioned conjugate gradients was the fastest. Gauss-Seidel with residuals update would be the method of choice for variance component estimation as well as solving. [ABSTRACT FROM AUTHOR]

The objective was to determine the effect of using core animals from different generations in single-step genomic BLUP with the Algorithm for Proven and Young (APY). Effective population size and number of independent chromosome segments (ICS) are limited in livestock populations, indicating limited dimensionality of genomic information. The APY takes advantage of this dimensionality and assumes that breeding vales (BV) for noncore animals are functions of the BV for core animals. The core animals represent the same information as the ICS. Simulations comprised a moderately heritable trait for 95,010 animals and 50,000 genotypes for animals across 5 generations. Genotypes consisted of 25,500 SNP distributed across 15 chromosomes. Core animals were defined based on individual generations, equal representation across generations, and at random. For a sufficiently large core size, core definitions had the same accuracies (r² = 0.90 ± 0.01) and biases (β1 = 1.02 ± 0.01) for young animals, even if the core animals had imperfect genotypes because of imputation. Using the youngest generations as core caused an increase in the number of rounds to convergence indicating some numerical instability with these core definitions. When 80% of genotyped animals had unknown parents, accuracy and bias were significantly better (P ≤ 0.05) for random and across-generation core definitions (r² = 0.71 ± 0.01; β = 0.75 ± 0.01) than for single-generation core definitions (r² = 0.61 ± 0.01; β = 0.53 ± 0.01). This difference could result from improved relationship estimates between animals in different generations because all generations were represented in the core partition that was directly inverted in APY. Thus, any subset of genotyped animals can be used to approximate the ICS when pedigrees are complete, but core animals should represent all generations when pedigrees are incomplete. [ABSTRACT FROM AUTHOR]

Single-step genomic BLUP (ssGBLUP) requires compatibility between genomic and pedigree relationships for unbiased and accurate predictions. Scaling the genomic relationship matrix (G) to have the same averages as the pedigree relationship matrix (i.e., scaling by averages) is one way to ensure compatibility. This requires computing both relationship matrices, calculating averages, and changing G , whereas only the inverses of those matrices are needed in the mixed model equations. Therefore, the compatibility process can add extra computing burden. In the single-step Bayesian regression, the scaling is done by including a mean (μ g) as a fixed effect in the model. The parameter μ g can be interpreted as the average of the breeding values of the genotyped animals. In this study, such scaling, called automatic, was implemented in ssGBLUP via Quaas-Pollak transformation of the inverse of the relationship matrix used in ssGBLUP (H), which combines the inverses of the pedigree and genomic relationship matrices. Comparisons involved a simulated data set, and the genomic relationship matrix was computed using different allele frequencies either from the current population (i.e., realized allele frequencies), equal among all the loci, or from the base population. For all of the scenarios, we computed bias [defined as the average difference between true breeding values (TBV) and genomic estimated breeding values (GEBV)], accuracy (defined as the correlation between TBV and GEBV), and dispersion (defined as the regression coefficient of GEBV on TBV). With no scaling, the bias expressed in terms of genetic standard deviations was 0.86, 0.64, and 0.58 with realized, equal, and base population allele frequencies, respectively. With scaling by averages, which is currently used in ssGBLUP, bias was 0.07, 0.08, and 0.03, respectively. With automatic scaling, bias was 0.18 regardless of allele frequencies. Accuracies were similar among scaling methods, but about 0.1 lower in the scenario without scaling. The GEBV were more inflated without any scaling, whereas the automatic scaling performed similarly to the scaling by averages. The average dispersion for those methods was 0.94. When μ g was treated as random, with the variance equal to differences between pedigree and genomic relationships, the bias was the same as with the scaling by averages. The automatic scaling is biased, especially when μ g is treated as a fixed effect. The bias may be small in real data with fewer generations, when traits are undergoing weak selection, or when the number of genotyped animals is large. [ABSTRACT FROM AUTHOR]

Genomic selection (GS) was applied to all major farm species including dairy, beef, pigs, and chickens. Initial methodology was a multistep procedure consisting of BLUP evaluation, extraction of pseudo-observations for genotyped animals, running genomic predictions via estimation of SNP effects or genomic relationships, and combining genomic and conventional predictions in an index. A newer single-step procedure works as BLUP with combined pedigree and genomic relationships. Experiences with GS indicate that 1) a few thousand genotyped animals are required for a reasonable increase in accuracy over BLUP, 2) SNP effects are not QTL, 3) SNP selection seem to increase the accuracy of genomic EBV more with small but almost none with large genotyped population 4) differences between methods based on estimation of SNP effects and genomic relationships are small, 5) gains with SNP chip > 50,000 are minimal, 6) gains from genotyping low-accuracy animals are small, 7) across-breed/line prediction does not work, 8) genomic predictions decay over time requiring constant phenotyping, and 9) a single-step procedure is much simpler to use than a multistep procedure and is usually more accurate. Experiences are in agreement with findings that the number of effective SNP is a function of effective population size and equals about 10,000 for cattle and less than 5000 for pigs and broiler chicken. Gains in accuracy due to genomics for different types of animals can be explained by decomposition of genomic EBV due to phenotypes, pedigree, and genomic information. Practical applications of GS indicate many potential problems and potentially genomic prediction that is less accurate than the conventional prediction. Problems include genotyping errors, pedigree errors, incorrect imputation, issues with deregression, inaccurate index, etc. Any application of GS requires a proper validation; a validation technique used in dairy may be unsuitable for pigs. For commercial use, the most flexible yet least trouble-prone methodology is a single-step procedure that, with recently developed algorithm for inversion of genomic relationship matrix, can accommodate any number of genotyped animals at low cost. [ABSTRACT FROM AUTHOR]

Official multibreed genomic evaluations for dairy cattle in the United States are based on multibreed BLUP evaluation followed by single-breed estimation of SNP effects. Single-step genomic BLUP (ssGBLUP) allows the straight computation of genomic (G)EBV in a multibreed context. This work aimed to develop ssGBLUP multibreed genomic predictions for US dairy cattle using the algorithm for proven and young (APY) to compute the inverse of the genomic relationship matrix. Only purebred Ayrshire (AY), Brown Swiss (BS), Guernsey (GU), Holstein (HO), and Jersey (JE) animals were considered. A 3-trait model with milk (MY), fat (FY), and protein (PY) yields was applied using about 45 million phenotypes recorded from January 2000 to June 2020. The whole data set included about 29.5 million animals, of which almost 4 million were genotyped. All the effects in the model were breed specific, and breed was also considered as fixed unknown parent groups. Evaluations were done for (1) each single breed separately (single); (2) HO and JE together (HO_JE); (3) AY, BS, and GU together (AY_BS_GU); (4) all the 5 breeds together (5_BREEDS). Initially, 15k core animals were used in APY for AY_BS_GU and 5_BREEDS, but larger core sets with more animals from the least represented breeds were also tested. The HO_JE evaluation had a fixed set of 30k core animals, with an equal representation of the 2 breeds, whereas HO and JE single-breed analysis involved 15k core animals. Validation for cows was based on correlations between adjusted phenotypes and (G)EBV, whereas for bulls on the regression of daughter yield deviations on (G)EBV. Because breed was correctly considered in the model, BLUP results for single and multibreed analyses were the same. Under ssGBLUP, predictability and reliability for AY, BS, and GU were on average 7% and 2% lower in 5_BREEDS compared with single-breed evaluations, respectively. However, validation parameters for these 3 breeds became better than in the single-breed evaluations when 45k animals were included in the core set for 5_BREEDS. Evaluations for Holsteins were more stable across scenarios because of the greatest number of genotyped animals and amount of data. Combining AY, BS, and GU into one evaluation resulted in predictions similar to the ones from single breed, especially when using about 30k core animals in APY. The results showed that single-step large-scale multibreed evaluations are computationally feasible, but fine tuning is needed to avoid a reduction in reliability when numerically dominant breeds are combined. Having evaluations for AY, BS, and GU separated from HO and JE may reduce inflation of GEBV for the first 3 breeds. [ABSTRACT FROM AUTHOR]

Some genomic evaluation models require creation and inversion of a genomic relationship matrix (G). As the number of genotyped animals increases, G becomes larger and thus requires more time for inversion. A single-step genomic evaluation also requires inversion of the part of the pedigree relationship matrix for genotyped animals (A22). A strategy was developed to provide an approximation of the inverse of G (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed. ) that may also be applied to the inverse of A22 ( Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed. ). The algorithm proceeds by creation of an incomplete Cholesky factorization (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed.) of G-1. For this purpose, a genomic relationship threshold determines whether 2 animals are closely related. For any animal, the sparsity pattern of the corresponding line in (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed. ) will thus gather elements corresponding to all close relatives of that animal. Any line of (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed.) is filled in with resulting estimators of the least-squares regression of genomic relationships between close relatives on genomic relationship between the animal considered and those close relatives. The (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed.) was computed as the matrix product (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed.) , where D-1 is a diagonal matrix. Then, (Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed. ) resulted in a new matrix that is close to diagonal and also needs to be inverted. The inverse of that matrix was approximated with the same decomposition as for approximation of the inverse of G ( Due to image rights restrictions, multiple line equation(s) cannot be graphically displayed. ) , and the procedure was repeated in successive rounds of recursion until a matrix was obtained that was close enough to diagonal to be inverted element by element. Two applications of the approximation algorithm were tested in a single-step genomic evaluation of US Holstein final score, and correlation coefficients between estimated breeding values based on either real or approximated G-1 were compared. Approximations came closer to G-1 as the number of recursion rounds increased. Approximations were even more accurate and expected to be faster for A22. Timesaving strategies are needed to reduce the computing time required for the algorithm. [ABSTRACT FROM AUTHOR]

Dense molecular markers are being used in genetic evaluation for parts of the population. This requires a two-step procedure where pseudo-data (for instance, daughter yield deviations) are computed from full records and pedigree data and later used for genomic evaluation. This results in bias and loss of information. One way to incorporate the genomic information into a full genetic evaluation is by modifying the numerator relationship matrix. A naive proposal is to substitute the relationships of genotyped animals with the genomic relationship matrix. However, this results in incoherencies because the genomic relationship matrix includes information on relationships among ancestors and descendants. In other words, using the pedigree-derived covariance between genotyped and ungenotyped individuals, with the pretense that genomic information does not exist, leads to inconsistencies. It is proposed to condition the genetic value of ungenotyped animals on the genetic value of genotyped animals via the selection index (e.g., pedigree information), and then use the genomic relationship matrix for the latter. This results in a joint distribution of genotyped and ungenotyped genetic values, with a pedigree-genomic relationship matrix H. In this matrix, genomic information is transmitted to the covariances among all ungenotyped individuals. The matrix is (semi)positive definite by construction, which is not the case for the naive approach. Numerical examples and alternative expressions are discussed. Matrix H is suitable for iteration on data algorithms that multiply a vector times a matrix, such as preconditioned conjugated gradients. [ABSTRACT FROM AUTHOR]

The objective of this study was to assess the reliability and bias of estimated breeding values (EBV) from traditional BLUP with unknown parent groups (UPG), genomic EBV (GEBV) from single-step genomic BLUP (ssGBLUP) with UPG for the pedigree relationship matrix (A) only (SS_UPG), and GEBV from ssGBLUP with UPG for both A and the relationship matrix among genotyped animals (A 22 ; SS_UPG2) using 6 large phenotype-pedigree truncated Holstein data sets. The complete data included 80 million records for milk, fat, and protein yields from 31 million cows recorded since 1980. Phenotype-pedigree truncation scenarios included truncation of phenotypes for cows recorded before 1990 and 2000 combined with truncation of pedigree information after 2 or 3 ancestral generations. A total of 861,525 genotyped bulls with progeny and cows with phenotypic records were used in the analyses. Reliability and bias (inflation/deflation) of GEBV were obtained for 2,710 bulls based on deregressed proofs, and on 381,779 cows born after 2014 based on predictivity (adjusted cow phenotypes). The BLUP reliabilities for young bulls varied from 0.29 to 0.30 across traits and were unaffected by data truncation and number of generations in the pedigree. Reliabilities ranged from 0.54 to 0.69 for SS_UPG and were slightly affected by phenotype-pedigree truncation. Reliabilities ranged from 0.69 to 0.73 for SS_UPG2 and were unaffected by phenotype-pedigree truncation. The regression coefficient of bull deregressed proofs on (G)EBV (i.e., GEBV and EBV) ranged from 0.86 to 0.90 for BLUP, from 0.77 to 0.94 for SS_UPG, and was 1.00 ± 0.03 for SS_UPG2. Cow predictivity ranged from 0.22 to 0.28 for BLUP, 0.48 to 0.51 for SS_UPG, and 0.51 to 0.54 for SS_UPG2. The highest cow predictivities for BLUP were obtained with the most extreme truncation, whereas for SS_UPG2, cow predictivities were also unaffected by phenotype-pedigree truncations. The regression coefficient of cow predictivities on (G)EBV was 1.02 ± 0.02 for SS_UPG2 with the most extreme truncation, which indicated the least biased predictions. Computations with the complete data set took 17 h with BLUP, 58 h with SS_UPG, and 23 h with SS_UPG2. The same computations with the most extreme phenotype-pedigree truncation took 7, 36, and 15 h, respectively. The SS_UPG2 converged in fewer rounds than BLUP, whereas SS_UPG took up to twice as many rounds. Thus, the ssGBLUP with UPG assigned to both A and A 22 provided accurate and unbiased evaluations, regardless of phenotype-pedigree truncation scenario. Old phenotypes (before 2000 in this data set) did not affect the reliability of predictions for young selection candidates, especially in SS_UPG2. [ABSTRACT FROM AUTHOR]

The purpose of this study was to evaluate the accuracy of genomic selection in single-step genomic BLUP (ssGBLUP) when the inverse of the genomic relationship matrix (G) is derived by the "algorithm for proven and young animals" (APY). This algorithm implements genomic recursions on a subset of "proven" animals. Only a relationship matrix for animals treated as "proven" needs to be inverted, and the extra costs of adding animals treated as "young" are linear. Analyses involved 10,102,702 final scores on 6,930,618 Holstein cows. Final score, which is a composite of type traits, is popular trait in the United States and was easily available for this study. A total of 100,000 animals with genotypes were used in the analyses and included 23,000 sires (16,000 with >5 progeny), 27,000 cows, and 50,000 young animals. Genomic EBV (GEBV) were calculated with a regular inverse of G, and with the G inverse approximated by APY. Animals in the proven subset included only sires (23,000), sires + cows (50,000), only cows (27,000), or sires with >5 progeny (16,000). The correlations of GEBV with APY and regular GEBV for young genotyped animals were 0.994, 0.995, 0.992, and 0.992, respectively Later, animals in the proven subset were randomly sampled from all genotyped animals in sets of 2,000, 5,000, 10,000, 15,000, and 20,000; each sample was replicated 4 times. Respective correlations were 0.97 (5,000 sample), 0.98 (10,000 sample), and 0.99 (20,000 sample), with minimal difference between samples of the same size. Genomic EBV with APY were accurate when the number of animals used in the subset is between 10,000 and 20,000, with little difference between the ways of creating the subset. Due to the approximately linear cost of APY, ssGBLUP with APY could support any number of genotyped animals without affecting accuracy. [ABSTRACT FROM AUTHOR]