Malena Erbe

Genotype x environment (GxE) interaction for production traits in Fleckvieh cattle was assessed by means of various environmental descriptors (EDs). It was also of particular interest to search for EDs useful for studying differing robustness or resilience of individuals which implies reasonable GxE interaction. The set of studied EDs included farm/herd environment (e.g. herd size, housing/feeding regimes, herd production level), geographical location (e.g. height above sea level), temperature humidity index and fat-to-protein ratio. Milk, fat and protein yield deviations for the first test day, the first 100 days and the 305-day-yield of the first lactation were used as phenotypes. Genetic correlations were estimated with bi- and multivariate sire models in case of categorical EDs and reaction norm sire models in case of continuous EDs. Further, rank correlations of bulls' estimated breeding values were calculated for different environmental levels/classes. Almost all estimated genetic correlations were significantly different from 1, but not <0.93. There were some exceptions for extremely different classes/levels of some EDs (e.g. average herd-year production, organic vs conventional farm systems, way of concentrated feed supply). Rank correlations were rarely below 0.95. In summary, no substantial GxE interactions for milk production traits were found with the studied EDs.

Haplotype data contains more information than genotype data and provides possibilities such as imputing low frequency variants, inferring points of recombination, detecting recurrent mutations, mapping linkage disequilibrium (LD), studying selection signatures, estimating IBD probabilities, etc. In addition, haplotype structure is used to assess genetic diversity and expected accuracy in genomic selection programs. Nevertheless, the quality and efficiency of phasing has rarely been a subject of thorough study but was assessed mainly as a by-product in imputation quality studies. Moreover, phasing studies based on data of a poultry population are non-existent. The aim of this study was to evaluate the phasing quality of FImpute and Beagle, two of the most used phasing software. We simulated ten replicated samples of a layer population comprising 888 individuals from a real SNP dataset of 580 k and a pedigree of 12 generations. Chromosomes analyzed were 1, 7 and 20. We measured the percentage of SNPs that were phased equally between true and phased haplotypes (Eqp), proportion of individuals completely correctly phased, number of incorrectly phased SNPs or Breakpoints (Bkp) and the length of inverted haplotype segments. Results were obtained for three different groups of individuals, with no parents or offspring genotyped in the dataset, with only one parent, and with both parents, respectively. The phasing was performed with Beagle (v3.3 and v4.1) and FImpute v2.2 (with and without pedigree). Eqp values ranged from 88 to 100%, with the best results from haplotypes phased with Beagle v4.1 and FImpute with pedigree information and at least one parent genotyped. FImpute haplotypes showed a higher number of Bkp than Beagle. As a consequence, switched haplotype segments were longer for Beagle than for FImpute. We concluded that for the dataset applied in this study Beagle v4.1 or FImpute with pedigree information and at least one parent genotyped in the data set were the best alternatives for obtaining high quality phased haplotypes.

This study aimed at (1) assessing the genomic stratification of experimental lines of Nelore cattle that have experienced different selection regimes for growth traits, and (2) identifying genomic regions that have undergone recent selection. We used a sample of 763 animals genotyped with the Illumina BovineHD BeadChip, among which 674 animals originated from two lines that are maintained under directional selection for increased yearling body weight and 89 animals from a control line that is maintained under stabilizing selection. Multidimensional analysis of the genomic dissimilarity matrix and admixture analysis revealed a substantial level of population stratification between the directional selection lines and the stabilizing selection control line. Two of the three tests used to detect selection signatures (FST, XP-EHH and iHS) revealed six candidate regions with indications of selection, which strongly indicates truly positive signals. The set of identified candidate genes included several genes with roles that are functionally related to growth metabolism, such as COL14A1, CPT1C, CRH, TBC1D1, and XKR4. The current study identified genetic stratification that resulted from almost four decades of divergent selection in an experimental Nelore population, and highlighted autosomal genomic regions that present patterns of recent selection. Our findings provide a basis for a better understanding of the metabolic mechanism that underlies the growth traits, which are modified by selection for yearling body weight.

Important traits in agricultural, natural, and human populations are increasingly being shown to be under the control of many genes that individually contribute only a small proportion of genetic variation. However, the majority of modern tools in quantitative and population genetics, including genome wide association studies and selection mapping protocols, are designed to identify individual genes with large effects. We have developed an approach to identify traits that have been under selection and are controlled by large numbers of loci. In contrast to existing methods, our technique utilizes additive effects estimates from all available markers, and relates these estimates to allele frequency change over time. Using this information, we generate a composite statistic, denoted Ĝ , which can be used to test for significant evidence of selection on a trait. Our test requires pre- and post-selection genotypic data but only a single time point with phenotypic information. Simulations demonstrate that Ĝ is powerful for identifying selection, particularly in situations where the trait being tested is controlled by many genes, which is precisely the scenario where classical approaches for selection mapping are least powerful. We apply this test to breeding populations of maize and chickens, where we demonstrate the successful identification of selection on traits that are documented to have been under selection.

Summary: Selection decisions in genomic selection schemes are made based on genomic breeding values (GBV) of candidates. Thus, the accuracy of GBV is a relevant parameter, as it reflects the stability of prediction and the possibility that the GBV might change when more information becomes available. Accuracy of genomic prediction defined as the correlation between GBV and true breeding values (TBV), however, is difficult to assess, considering TBV of the candidates are not available in reality. In previous studies, several methods were proposed to assess the accuracy of GBV including methods using population parameters or parameters inferred from mixed‐model equations. In practice, most approaches tended to overestimate the accuracy of genomic prediction. We thus tested approaches used in previous studies in order to assess the magnitude of bias. Analyses were performed based on simulated data under a variety of scenarios mimicking different livestock breeding programmes. Furthermore, we proposed a novel method and tested it both with simulated data and in a real Holstein data set. The new method provided a better prediction for the accuracy of GBV in the simulated scenarios.

With the availability of next-generation sequencing technologies, genomic prediction based on whole-genome sequencing (WGS) data is now feasible in animal breeding schemes and was expected to lead to higher predictive ability, since such data may contain all genomic variants including causal mutations. Our objective was to compare prediction ability with high-density (HD) array data and WGS data in a commercial brown layer line with genomic best linear unbiased prediction (GBLUP) models using various approaches to weight single nucleotide polymorphisms (SNPs). A total of 892 chickens from a commercial brown layer line were genotyped with 336 K segregating SNPs (array data) that included 157 K genic SNPs (i.e. SNPs in or around a gene). For these individuals, genome-wide sequence information was imputed based on data from re-sequencing runs of 25 individuals, leading to 5.2 million (M) imputed SNPs (WGS data), including 2.6 M genic SNPs. De-regressed proofs (DRP) for eggshell strength, feed intake and laying rate were used as quasi-phenotypic data in genomic prediction analyses. Four weighting factors for building a trait-specific genomic relationship matrix were investigated: identical weights, −(log10 P) from genome-wide association study results, squares of SNP effects from random regression BLUP, and variable selection based weights (known as BLUP|GA). Predictive ability was measured as the correlation between DRP and direct genomic breeding values in five replications of a fivefold cross-validation. Averaged over the three traits, the highest predictive ability (0.366 ± 0.075) was obtained when only genic SNPs from WGS data were used. Predictive abilities with genic SNPs and all SNPs from HD array data were 0.361 ± 0.072 and 0.353 ± 0.074, respectively. Prediction with −(log10 P) or squares of SNP effects as weighting factors for building a genomic relationship matrix or BLUP|GA did not increase accuracy, compared to that with identical weights, regardless of the SNP set used. Our results show that little or no benefit was gained when using all imputed WGS data to perform genomic prediction compared to using HD array data regardless of the weighting factors tested. However, using only genic SNPs from WGS data had a positive effect on prediction ability.

Epistasis marker effect models incorporating products of marker values as predictor variables in a linear regression approach (extended GBLUP, EGBLUP) have been assessed as potentially beneficial for genomic prediction, but their performance depends on marker coding. Although this fact has been recognized in literature, the nature of the problem has not been thoroughly investigated so far. We illustrate how the choice of marker coding implicitly specifies the model of how effects of certain allele combinations at different loci contribute to the phenotype, and investigate coding-dependent properties of EGBLUP. Moreover, we discuss an alternative categorical epistasis model (CE) eliminating undesired properties of EGBLUP and show that the CE model can improve predictive ability. Finally, we demonstrate that the coding-dependent performance of EGBLUP offers the possibility to incorporate prior experimental information into the prediction method by adapting the coding to already available phenotypic records on other traits. Based on our results, for EGBLUP, a symmetric coding {−1,1} or {−1,0,1} should be preferred, whereas a standardization using allele frequencies should be avoided. Moreover, CE can be a valuable alternative since it does not possess the undesired theoretical properties of EGBLUP. However, which model performs best will depend on characteristics of the data and available prior information. Data from previous experiments can for instance be incorporated into the marker coding of EGBLUP.

The technical progress in the last decade has made it possible to sequence millions of DNA reads in a relatively short time frame. Several variant callers based on different algorithms have emerged and have made it possible to extract single nucleotide polymorphisms (SNPs) out of the whole-genome sequence. Often, only a few individuals of a population are sequenced completely and imputation is used to obtain genotypes for all sequence-based SNP loci for other individuals, which have been genotyped for a subset of SNPs using a genotyping array. First, we compared the sets of variants detected with different variant callers, namely GATK, freebayes and SAMtools, and checked the quality of genotypes of the called variants in a set of 50 fully sequenced white and brown layers. Second, we assessed the imputation accuracy (measured as the correlation between imputed and true genotype per SNP and per individual, and genotype conflict between father-progeny pairs) when imputing from high density SNP array data to whole-genome sequence using data from around 1000 individuals from six different generations. Three different imputation programs (Minimac, FImpute and IMPUTE2) were checked in different validation scenarios. There were 1,741,573 SNPs detected by all three callers on the studied chromosomes 3, 6, and 28, which was 71.6 % (81.6 %, 88.0 %) of SNPs detected by GATK (SAMtools, freebayes) in total. Genotype concordance (GC) defined as the proportion of individuals whose array-derived genotypes are the same as the sequence-derived genotypes over all non-missing SNPs on the array were 0.98 (GATK), 0.97 (freebayes) and 0.98 (SAMtools). Furthermore, the percentage of variants that had high values (>0.9) for another three measures (non-reference sensitivity, non-reference genotype concordance and precision) were 90 (88, 75) for GATK (SAMtools, freebayes). With all imputation programs, correlation between original and imputed genotypes was >0.95 on average with randomly masked 1000 SNPs from the SNP array and >0.85 for a leave-one-out cross-validation within sequenced individuals. Performance of all variant callers studied was very good in general, particularly for GATK and SAMtools. FImpute performed slightly worse than Minimac and IMPUTE2 in terms of genotype correlation, especially for SNPs with low minor allele frequency, while it had lowest numbers in Mendelian conflicts in available father-progeny pairs. Correlations of real and imputed genotypes remained constantly high even if individuals to be imputed were several generations away from the sequenced individuals.