Ricardo Pong-Wong

The Impact of Genetic Architecture on Genome-Wide Evaluation Methods

The rapid increase in high-throughput single-nucleotide polymorphism data has led to a great interest in applying genome-wide evaluation methods to identify an individuals genetic merit. Genome-wide evaluation combines statistical methods with genomic data to predict genetic values for complex traits. Considerable uncertainty currently exists in determining which genome-wide evaluation method is the most appropriate. We hypothesize that genome-wide methods deal differently with the genetic architecture of quantitative traits and genomes. A genomic linear method (GBLUP), and a genomic nonlinear Bayesian variable selection method (BayesB) are compared using stochastic simulation across three effective population sizes and a wide range of numbers of quantitative trait loci (NQTL). GBLUP had a constant accuracy, for a given heritability and sample size, regardless of NQTL. BayesB had a higher accuracy than GBLUP when NQTL was low, but this advantage diminished as NQTL increased and when NQTL became large, GBLUP slightly outperformed BayesB. In addition, deterministic equations are extended to predict the accuracy of both methods and to estimate the number of independent chromosome segments (Me) and NQTL. The predictions of accuracy and estimates of Me and NQTL were generally in good agreement with results from simulated data. We conclude that the relative accuracy of GBLUP and BayesB for a given number of records and heritability are highly dependent on Me, which is a property of the target genome, as well as the architecture of the trait (NQTL).

Whole-Genome Regression and Prediction Methods Applied to Plant and Animal Breeding

Genomic-enabled prediction is becoming increasingly important in animal and plant breeding and is also receiving attention in human genetics. Deriving accurate predictions of complex traits requires implementing whole-genome regression (WGR) models where phenotypes are regressed on thousands of markers concurrently. Methods exist that allow implementing these large-p with small-n regressions, and genome-enabled selection (GS) is being implemented in several plant and animal breeding programs. The list of available methods is long, and the relationships between them have not been fully addressed. In this article we provide an overview of available methods for implementing parametric WGR models, discuss selected topics that emerge in applications, and present a general discussion of lessons learned from simulation and empirical data analysis in the last decade.

Genomic Prediction in Animals and Plants: Simulation of Data, Validation, Reporting and Benchmarking

The genomic prediction of phenotypes and breeding values in animals and plants has developed rapidly into its own research field. Results of genomic prediction studies are often difficult to compare because data simulation varies, real or simulated data are not fully described, and not all relevant results are reported. In addition, some new methods have been compared only in limited genetic architectures, leading to potentially misleading conclusions. In this article we review simulation procedures, discuss validation and reporting of results, and apply benchmark procedures for a variety of genomic prediction methods in simulated and real example data. Plant and animal breeding programs are being transformed by the use of genomic data, which are becoming widely available and cost-effective to predict genetic merit. A large number of genomic prediction studies have been published using both simulated and real data. The relative novelty of this area of research has made the development of scientific conventions difficult with regard to description of the real data, simulation of genomes, validation and reporting of results, and forward in time methods. In this review article we discuss the generation of simulated genotype and phenotype data, using approaches such as the coalescent and forward in time simulation. We outline ways to validate simulated data and genomic prediction results, including cross-validation. The accuracy and bias of genomic prediction are highlighted as performance indicators that should be reported. We suggest that a measure of relatedness between the reference and validation individuals be reported, as its impact on the accuracy of genomic prediction is substantial. A large number of methods were compared in example simulated and real (pine and wheat) data sets, all of which are publicly available. In our limited simulations, most methods performed similarly in traits with a large number of quantitative trait loci (QTL), whereas in traits with fewer QTL variable selection did have some advantages. In the real data sets examined here all methods had very similar accuracies. We conclude that no single method can serve as a benchmark for genomic prediction. We recommend comparing accuracy and bias of new methods to results from genomic best linear prediction and a variable selection approach (e.g., BayesB), because, together, these methods are appropriate for a range of genetic architectures. An accompanying article in this issue provides a comprehensive review of genomic prediction methods and discusses a selection of topics related to application of genomic prediction in plants and animals.

A simple and rapid method for calculating identity-by-descent matrices using multiple markers

A fast, partly recursive deterministic method for calculating Identity-by-Descent (IBD) probabilities was developed with the objective of using IBD in Quantitative Trait Locus (QTL) mapping. The method combined a recursive method for a single marker locus with a method to estimate IBD between sibs using multiple markers. Simulated data was used to compare the deterministic method developed in the present paper with a stochastic method (LOKI) for precision in estimating IBD probabilities and performance in the task of QTL detection with the variance component approach. This comparison was made in a variety of situations by varying family size and degree of polymorphism among marker loci. The following were observed for the deterministic method relative to MCMC: (i) it was an order of magnitude faster; (ii) its estimates of IBD probabilities were found to agree closely, even though it does not extract information when haplotypes are not known with certainty; (iii) the shape of the profile for the QTL test statistic as a function of location was similar, although the magnitude of the test statistic was slightly smaller; and (iv) the estimates of QTL variance was similar. It was concluded that the method proposed provided a rapid means of calculating the IBD matrix with only a small loss in precision, making it an attractive alternative to the use of stochastic MCMC methods. Furthermore, developments in marker technology providing denser maps would enhance the relative advantage of this method.

Homozygosity for a single base-pair mutation in the oocyte-specific GDF9 gene results in sterility in Thoka sheep

The control of fecundity is critical in determining mammalian offspring survival. It is regulated principally by the ovulation rate, so that primates and large farm species commonly have a single offspring. Previously, several mutations have been identified in sheep which increase the naturally low ovulation rate; although in some cases homozygous ewes are infertile. In the present study we present a detailed characterization of a novel mutation in growth differentiation factor 9 (GDF9), found in Icelandic Thoka sheep. This mutation is a single base change (A1279C) resulting in a nonconservative amino acid change (S109R) in the C-terminus of the mature GDF9 protein, which is normally expressed in oocytes at all stages of development. Genotyping all animals for which reproductive records were available confirmed this mutation to be associated with increased fecundity in heterozygous ewes and infertility in homozygotes. Analysis of homozygote ovarian morphology and a number of genes normally activated in growing follicles showed that GDF9 was not involved in oocyte activation, but in subsequent development of the follicle. This study highlights the importance of oocyte factors in regulating fertility and provides new information for structural analysis and investigation of the potentially important sites of dimerization or translational modifications required to produce biologically active GDF9. It also provides the basis for the utilization of these animals to enhance sheep production.

Application of high-dimensional feature selection: evaluation for genomic prediction in man

In this study, we investigated the effect of five feature selection approaches on the performance of a mixed model (G-BLUP) and a Bayesian (Bayes C) prediction method. We predicted height, high density lipoprotein cholesterol (HDL) and body mass index (BMI) within 2,186 Croatian and into 810 UK individuals using genome-wide SNP data. Using all SNP information Bayes C and G-BLUP had similar predictive performance across all traits within the Croatian data, and for the highly polygenic traits height and BMI when predicting into the UK data. Bayes C outperformed G-BLUP in the prediction of HDL, which is influenced by loci of moderate size, in the UK data. Supervised feature selection of a SNP subset in the G-BLUP framework provided a flexible, generalisable and computationally efficient alternative to Bayes C; but careful evaluation of predictive performance is required when supervised feature selection has been used.

Genome-wide association study identifies novel loci associated with resistance to bovine tuberculosis

Tuberculosis (TB) caused by Mycobacterium bovis is a re-emerging disease of livestock that is of major economic importance worldwide, as well as being a zoonotic risk. There is significant heritability for host resistance to bovine TB (bTB) in dairy cattle. To identify resistance loci for bTB, we undertook a genome-wide association study in female Holstein–Friesian cattle with 592 cases and 559 age-matched controls from case herds. Cases and controls were categorised into distinct phenotypes: skin test and lesion positive vs skin test negative on multiple occasions, respectively. These animals were genotyped with the Illumina BovineHD 700K BeadChip. Genome-wide rapid association using linear and logistic mixed models and regression (GRAMMAR), regional heritability mapping (RHM) and haplotype-sharing analysis identified two novel resistance loci that attained chromosome-wise significance, protein tyrosine phosphatase receptor T (PTPRT; P=4.8 × 10−7) and myosin IIIB (MYO3B; P=5.4 × 10−6). We estimated that 21% of the phenotypic variance in TB resistance could be explained by all of the informative single-nucleotide polymorphisms, of which the region encompassing the PTPRT gene accounted for 6.2% of the variance and a further 3.6% was associated with a putative copy number variant in MYO3B. The results from this study add to our understanding of variation in host control of infection and suggest that genetic marker-based selection for resistance to bTB has the potential to make a significant contribution to bTB control.

Impact of biotechnology on (cross)breeding programmes in pigs.

Abstract Crossbreeding programmes in pigs exploit between breed complementarity of additive genetic effects and heterosis generated by non-additive genetic effects. Within breed, improvement programmes may focus on additive effects and hence the enhancement of complementarity, but non-additive variation is not generally used in within line selection or for mate selection at the multiplier or commercial level. In this paper, we discuss the impact of new biotechnological tools, particularly molecular markers, multiple ovulation and embryo transfer (MOET), and cloning, on structures and methods in crossbreeding. At the between line level, genetic marker information could allow better prediction of heterosis in novel crosses from information on genetic distances. Within the crossbreeding structure, the same technique might be applied at the multiplier and commercial level to exploit specific combining abilities of particular animals. Combining simple MOET and cloning protocols could radically alter the dissemination of crossbreeding benefits and their delivery to the farmer. The combination of MOET, cloning and genomic tools could result in speed genetics programmes, i.e. fast introgression and recurrent selection methods. Thus, the ultimate impact of biotechnology will be increased rates of progress, efficient use of variation, reduced genetic lag, and the removal of one or two tiers in the breeding pyramid. The costs of new technologies are discussed briefly.

Genome-wide association and regional heritability mapping to identify loci underlying variation in nematode resistance and body weight in Scottish Blackface lambs

The genetic architecture underlying nematode resistance and body weight in Blackface lambs was evaluated comparing genome-wide association (GWA) and regional heritability mapping (RHM) approaches. The traits analysed were faecal egg count (FEC) and immunoglobulin A activity against third-stage larvae from Teladorsagia circumcincta, as indicators of nematode resistance, and body weight in a population of 752 Scottish Blackface lambs, genotyped with the 50k single-nucleotide polymorphism (SNP) chip. FEC for both Nematodirus and Strongyles nematodes (excluding Nematodirus), as well as body weight were collected at approximately 16, 20 and 24 weeks of age. In addition, a weighted average animal effect was estimated for both FEC and body weight traits. After quality control, 44 388 SNPs were available for the GWA analysis and 42 841 for the RHM, which utilises only mapped SNPs. The same fixed effects were used in both analyses: sex, year, management group, litter size and age of dam, with day of birth as covariate. Some genomic regions of interest for both nematode resistance and body weight traits were identified, using both GWA and RHM approaches. For both methods, strong evidence for association was found on chromosome 14 for Nematodirus average animal effect, chromosome 6 for Strongyles FEC at 16 weeks and chromosome 6 for body weight at 16 weeks. Across the entire data set, RHM identified more regions reaching the suggestive level than GWA, suggesting that RHM is capable of capturing some of the variation not detected by GWA analyses.

Response to mass selection when an identified major gene is segregating

A deterministic model to predict response with mass selection when a major locus is segregating is presented. The model uses a selection index framework in which the weight of the different components included in the index are adjusted to describe the different methods of selection using genotype information as selection criteria. The response over multiple generations to several methods of selection using either the whole genotype effect (genotypic methods) or only the Mendelian sampling deviation of the major locus (Mendelian methods) was compared with selection using only performance record (phenotypic method). Relevant differences in response between using and ignoring information on the major gene were observed only when the favourable allele was at a low frequency. When the major locus had a completely additive effect, all the genotypic or Mendelian methods had a higher cumulated genetic gain in the first 3-4 generations of selection but this advantage was lost thereafter. In the long term, without exception, all methods using genotype information of an additive major gene had lower cumulated gain than phenotypic selection over a wide range of parameters. The reason for the long-term loss, was a reduction in the intensity of selection applied to the polygenic background arising from increasing the differences in the selective advantage between genotype groups. The same trend was observed when the favourable allele of the major locus was completely recessive or dominant, with the exception of the cases of a large recessive locus (over one phenotypic standard deviation) where the extra early gain from using genotype information was maintained in the long term. This was explained by the inefficiency of the phenotypic selection to fix the favourable allele due to the linkage disequilibrium built-up between the major locus and the polygenic effects. Differences in the inbreeding rate were also observed between these methods: the genotypic methods had the highest inbreeding rate while the Mendelian had the lowest. The difference in the inbreeding rate was mainly observed in the first generations of selection and increased with lower starting frequency of the major locus. @ Inra/Elsevier, Paris major gene / indice / gain / inbreeding / loss