Deterministic predictions of the accuracy of genomic breeding values in selection candidates with no phenotypes have been derived based on the heritability of the trait, number of phenotyped and genotyped animals in the reference population where the marker effects are estimated, the effective population size and the length of the genome. We assessed the value of these deterministic predictions given the results that have been achieved in Holstein and Jersey dairy cattle. We conclude that the deterministic predictions are useful guide for establishing the size of the reference populations which must be assembled in order to predict genomic breeding values at a desired level of accuracy in selection candidates.

Since the 1990s many promises have been made about economic benefits available from genome scanning. We were assured that we would be able to look at an animal's genes and determine its genetic merit directly. Of course this included the assumption, generally implied, that we would have already determined the effects of all the (important) genes, at all times and under all circumstances. However, despite numerous 'in silico' proofs, it seems that ascertaining these effects is proving much more elusive than originally assumed. We did not comprehend how much data we would need. By mating the best to the best, we have been 'improving' domesticated species for millenia. We keep getting more sophisticated about determining what is 'best' but genetic improvement was achieved long before the mechanism of inheritance was understood. Despite shortcomings in our understanding of quantitative genetic variation (e.g. the search for the missing heritability), these methods are highly effective. Furthermore, these methods of genetic evaluation use no explicit knowledge of individual gene action. The question remains: 'Will such knowledge make predictions more accurate?'

A strategy to reduce computational demands of genome-wide association studies fitting a mixed model is presented. Improvements are achieved by utilizing a large proportion of calculations that remain constant across the multiple analyses for individual markers involved, with estimates obtained without inverting large matrices.

Whole genome association studies have been increasingly used for QTL discovery. Analyses can performed using single marker (for example, single-nucleotide polymorphism, SNP) regression, but have also evolved to multiple marker (SNPs) regression to facilitate genomic selection. A major area of importance to the Northern Australian beef industry is female reproductive rate; specifically the number of calves produced over the lifetime of breeding females. An important component of female reproductive rate is age at puberty of heifers. Reduction of age at puberty of heifers can increase calving rates (Taylor and Rudder 1986). This study compared the two types of methods for estimating SNP effects for age at puberty.

Genomic information is now commonly used in routine genetic evaluations. This is usually in the form of genomic breeding values (GBVs) which have a high heritability but are generally confined to those animals with genotypes. This can lead to anomalies when parents have GBVs and progeny do not. By using a single-trait genetic evaluation, GBVs can be generated for related individuals. It is most efficient to do this for genotyped individuals and their ancestors initially, and calculate mid-parent values for all other individuals. A method for approximating accuracies for the relatives' GBVs is described.

Background: The apparent effect of a single nucleotide polymorphism (SNP) on phenotype depends on the linkage disequilibrium (LD) between the SNP and a quantitative trait locus (QTL). However, the phase of LD between a SNP and a QTL may differ between 'Bos indicus' and 'Bos taurus' because they diverged at least one hundred thousand years ago. Here, we test the hypothesis that the apparent effect of a SNP on a quantitative trait depends on whether the SNP allele is inherited from a 'Bos taurus' or 'Bos indicus' ancestor. Methods: Phenotype data on one or more traits and SNP genotype data for 10 181 cattle from 'Bos taurus', 'Bos indicus' and composite breeds were used. All animals had genotypes for 729 068 SNPs (real or imputed). Chromosome segments were classified as originating from B. 'indicus' or B. 'taurus' on the basis of the haplotype of SNP alleles they contained. Consequently, SNP alleles were classified according to their sub-species origin. Three models were used for the association study: (1) conventional GWAS (genome-wide association study), fitting a single SNP effect regardless of subspecies origin, (2) interaction GWAS, fitting an interaction between SNP and subspecies-origin, and (3) best variable GWAS, fitting the most significant combination of SNP and sub-species origin. Results: Fitting an interaction between SNP and subspecies origin resulted in more significant SNPs (i.e. more power) than a conventional GWAS. Thus, the effect of a SNP depends on the subspecies that the allele originates from. Also, most QTL segregated in only one subspecies, suggesting that many mutations that affect the traits studied occurred after divergence of the subspecies or the mutation became fixed or was lost in one of the subspecies. Conclusions: The results imply that GWAS and genomic selection could gain power by distinguishing SNP alleles based on their subspecies origin, and that only few QTL segregate in both B. 'indicus' and B. 'taurus' cattle. Thus, the QTL that segregate in current populations likely resulted from mutations that occurred in one of the subspecies and can have both positive and negative effects on the traits. There was no evidence that selection has increased the frequency of alleles that increase body weight.

Background: The major obstacles for the implementation of genomic selection in Australian beef cattle are the variety of breeds and in general, small numbers of genotyped and phenotyped individuals per breed. The Australian Beef Cooperative Research Center (Beef CRC) investigated these issues by deriving genomic prediction equations (PE) from a training set of animals that covers a range of breeds and crosses including Angus, Murray Grey, Shorthorn, Hereford, Brahman, Belmont Red, Santa Gertrudis and Tropical Composite. This paper presents accuracies of genomically estimated breeding values (GEBV) that were calculated from these PE in the commercial pure-breed beef cattle seed stock sector. Methods: PE derived by the Beef CRC from multi-breed and pure-breed training populations were applied to genotyped Angus, Limousin and Brahman sires and young animals, but with no pure-breed Limousin in the training population. The accuracy of the resulting GEBV was assessed by their genetic correlation to their phenotypic target trait in a bi-variate REML approach that models GEBV as trait observations. Results: Accuracies of most GEBV for Angus and Brahman were between 0.1 and 0.4, with accuracies for abattoir carcass traits generally greater than for live animal body composition traits and reproduction traits. Estimated accuracies greater than 0.5 were only observed for Brahman abattoir carcass traits and for Angus carcass rib fat. Averaged across traits within breeds, accuracies of GEBV were highest when PE from the pooled across-breed training population were used. However, for the Angus and Brahman breeds the difference in accuracy from using pure-breed PE was small. For the Limousin breed no reasonable results could be achieved for any trait. Conclusion: Although accuracies were generally low compared to published accuracies estimated within breeds, they are in line with those derived in other multi-breed populations. Thus PE developed by the Beef CRC can contribute to the implementation of genomic selection in Australian beef cattle breeding.

A DNA test to assign poll genotype in Australian beef cattle breeds was released in 2010 and was based on a strong association between an allele (303) at marker CSAFG29 and the polled phenotype in Brahman and other tropical breeds. A pre-commercialisation field trial revealed that this marker was not able to accurately assign poll genotype in a number of other breeds, including Limousin and Brangus, due to the high frequency of another allele (305) that was known to be associated with both polled and horned across a variety of breeds. Using a haplotype test the current study demonstrates that there are two sources of the 305 allele in Limousin and several alleles at CSAFG29 that are associated with polled in Brangus. The haplotype based test has resulted in a much better diagnostic for polled in these breeds.

Use of high density Single Nucleotide Polymorphic (SNP) marker information allows for prediction of genetic merit via genome wide selection and for localization of markers in gene regions of biological interest through Genome Wide Association Studies (GWAS). We report on a replicated GWAS in dairy cattle using 1,945 progeny tested bulls genotyped with three high density SNP panels representing 63,678 informative SNP. Single SNP genotypes were analysed against deregressed EBV for protein percent and fat percent using a mixed linear model accounting for SNP and animal polygenic effects. The 127,356 analyses (63,678 informative SNP by two traits) across the two data sets identified 143 and 87 significant (P<0.05, corrected for False Discovery Rate) associations for protein % in data set 1 and 2 respectively, whilst for fat % 102 and 61 significant associations were identified in the two data sets respectively. Outputs from selected SNP analyses are discussed for significance and pleiotropic effects and compared against integrated QTL meta-assembly from public domain studies.

Background: The advent of genomic marker data has triggered the development of various Bayesian algorithms for estimation of marker effects, but software packages implementing these algorithms are not readily available, or are limited to a single algorithm, uni-variate analysis or a limited number of factors. Moreover, script based environments like R may not be able to handle large-scale genomic data or exploit model properties which save computing time or memory (RAM). Results: BESSiE is a software designed for best linear unbiased prediction (BLUP) and Bayesian Markov chain Monte Carlo analysis of linear mixed models allowing for continuous and/or categorical multivariate, repeated and missing observations, various random and fixed factors and large-scale genomic marker data. BESSiE covers the algorithms genomic BLUP, single nucleotide polymorphism (SNP)-BLUP, BayesA, BayesB, BayesCπ and BayesR for estimating marker effects and/or summarised genomic values. BESSiE is parameter file driven, command line operated and available for Linux environments. BESSiE executable, manual and a collection of examples can be downloaded http:// turing.une.edu.au/~agbu-admin/BESSiE/. Conclusion: BESSiE allows the user to compare several different Bayesian and BLUP algorithms for estimating marker effects from large data sets in complex models with the same software by small alterations in the parameter file. The program has no hard-coded limitations for number of factors, observations or genetic markers.