Experimental designs for reliable detection of linkage disequilibrium in unstructured random population association studies

Genome-Wide Association Study and Pathway Analysis for Female Fertility Traits in Iranian Holstein Cattle

Female fertility is an important trait that contributes to cow’s profitability and it can be improved by genomic information. The objective of this study was to detect genomic regions and variants affecting fertility traits in Iranian Holstein cattle. A data set comprised of female fertility records and 3,452,730 pedigree information from Iranian Holstein cattle were used to predict the breeding values, which were then employed to estimate the de-regressed proofs (DRP) of genotyped animals. A total of 878 animals with DRP records and 54k SNP markers were utilized in the genome-wide association study (GWAS). The GWAS was performed using a linear regression model with SNP genotype as a linear covariate. The results showed that an SNP on BTA19, ARS-BFGL-NGS-33473, was the most significant SNP associated with days from calving to first service. In total, [69] significant SNPs were located within 27 candidate genes. Novel potential candidate genes include OSTN, DPP6, EphA5, CADPS2, Rfc1, ADGRB3, Myo3a, C10H14orf93, KIAA1217, RBPJL, SLC18A2, GARNL3, NCALD, ASPH, ASIC2, OR3A1, CHRNB4, CACNA2D2, DLGAP1, GRIN2A and ME3. These genes are involved in different pathways relevant to female fertility and other characteristics in mammals. Gene set enrichment analysis showed that thirteen GO terms had significant overrepresentation of genes statistically associated with female fertility traits. The results of network analysis identified CCNB1 gene as a hub gene in the progesterone-mediated oocyte maturation pathway, significantly associated with age at first calving. The candidate genes identified in this study can be utilized in genomic tests to improve reproductive performance in Holstein cattle.

Building strong relationships between conservation genetics and primary industry leads to mutually beneficial genomic advances

Several reviews in the past decade have heralded the benefits of embracing high-throughput sequencing technologies to inform conservation policy and the management of threatened species, but few have offered practical advice on how to expedite the transition from conservation genetics to conservation genomics. Here, we argue that an effective and efficient way to navigate this transition is to capitalize on emerging synergies between conservation genetics and primary industry (e.g., agriculture, fisheries, forestry and horticulture). Here, we demonstrate how building strong relationships between conservation geneticists and primary industry scientists is leading to mutually-beneficial outcomes for both disciplines. Based on our collective experience as collaborative New Zealand-based scientists, we also provide insight for forging these cross-sector relationships.

The distribution of SNP marker effects for faecal worm egg count in sheep, and the feasibility of using these markers to predict genetic merit for resistance to worm infections

SummaryGenetic resistance to gastrointestinal worms is a complex trait of great importance in both livestock and humans. In order to gain insights into the genetic architecture of this trait, a mixed breed population of sheep was artificially infected with Trichostrongylus colubriformis (n=3326) and then Haemonchus contortus (n=2669) to measure faecal worm egg count (WEC). The population was genotyped with the Illumina OvineSNP50 BeadChip and 48 640 single nucleotide polymorphism (SNP) markers passed the quality controls. An independent population of 316 sires of mixed breeds with accurate estimated breeding values for WEC were genotyped for the same SNP to assess the results obtained from the first population. We used principal components from the genomic relationship matrix among genotyped individuals to account for population stratification, and a novel approach to directly account for the sampling error associated with each SNP marker regression. The largest marker effects were estimated to explain an average of 0·48% (T. colubriformis) or 0·08% (H. contortus) of the phenotypic variance in WEC. These effects are small but consistent with results from other complex traits. We also demonstrated that methods which use all markers simultaneously can successfully predict genetic merit for resistance to worms, despite the small effects of individual markers. Correlations of genomic predictions with breeding values of the industry sires reached a maximum of 0·32. We estimate that effective across-breed predictions of genetic merit with multi-breed populations will require an average marker spacing of approximately 10 kbp.

A Genome Wide Association Study on Age at First Calving Using High Density Single Nucleotide Polymorphism Chips in Hanwoo (Bos taurus coreanae)

Age at first calving is an important trait for achieving earlier reproductive performance. To detect quantitative trait loci (QTL) for reproductive traits, a genome wide association study was conducted on the 96 Hanwoo cows that were born between 2008 and 2010 from 13 sires in a local farm (Juk-Am Hanwoo farm, Suncheon, Korea) and genotyped with the Illumina 50K bovine single nucleotide polymorphism (SNP) chips. Phenotypes were regressed on additive and dominance effects for each SNP using a simple linear regression model after the effects of birth-year-month and polygenes were considered. A forward regression procedure was applied to determine the best set of SNPs for age at first calving. A total of 15 QTL were detected at the comparison-wise 0.001 level. Two QTL with strong statistical evidence were found at 128.9 Mb and 111.1 Mb on bovine chromosomes (BTA) 2 and 7, respectively, each of which accounted for 22% of the phenotypic variance. Also, five significant SNPs were detected on BTAs 10, 16, 20, 26, and 29. Multiple QTL were found on BTAs 1, 2, 7, and 14. The significant QTLs may be applied via marker assisted selection to increase rate of genetic gain for the trait, after validation tests in other Hanwoo cow populations.

Association of loblolly pine xylem development gene expression with single-nucleotide polymorphisms

Variation in the expression of genes with putative roles in wood development was associated with single-nucleotide polymorphisms (SNPs) using a population of loblolly pine (Pinus taeda L.) that included individuals from much of the native range. Association studies were performed using 3938 SNPs and expression data obtained using quantitative real-time polymerase chain reaction (PCR) (qRT-PCR) for 106 xylem development genes in 400 clonally replicated loblolly pine individuals. A general linear model (GLM) approach, which takes the underlying population structure into consideration, was used to discover significant associations. After adjustment for multiple testing using a false discovery rate correction, 88 statistically significant associations (Q<0.05) were observed for 80 SNPs with the expression data of 33 xylem development genes. Thirty SNPs caused nonsynonymous mutations, 18 resulted in synonymous mutations, 11 were in 3' untranslated regions (UTRs), 1 was in a 5' UTR and 20 were in introns. Using AraNet, we found that Arabidopsis genes with high similarity to the loblolly pine genes involved in 21 of the 88 statistically significant associations are connected in functional gene networks. Comparisons of gene expression values revealed that in most cases the average expression in plants homozygous for the rare SNP allele was lower than that of plants that were heterozygous or homozygous for the abundant allele. Although there are association studies of SNPs and expression profiles for humans, Arabidopsis and white spruce, to the best of our knowledge, this is the first example of such an association genetic study in pines. Functional validation of these associations will lead to a deeper understanding of the molecular basis of phenotypic differences in wood development among individuals in conifer populations.

Statistical analysis of genomic data

In this chapter we describe methods for statistical analysis of GWAS data with the goal of quantifying evidence for genomic effects associated with trait variation, while avoiding spurious associations due to evidence not being well quantified or due to population structure.Single marker analysis and imputation are discussed in Sect. 1, and a Bayesian multi-locus analysis using the BayesQTLBIC R package (1, 2) is described in Sect. 2. The multi-locus analysis, applied in a genomic window, enables local inference of the QTL genetic architecture and is an alternative to imputation. Multi-locus analysis with BayesQTLBIC, including calculation of posterior probabilities for alternative models, posterior probabilities for number of QTL, marginal probabilities for markers, and Bayes factors for individual chromosomes, is demonstrated for simulated QTL data. Methods for correcting the population structure and the possible effects of population structure on power are discussed in Sect. 3. Section 4 considers analysis combining information from linkage and linkage disequilibrium when sampling from a pedigree. Section 5 considers combining information from two different studies-showing that data from an existing QTL mapping family can be profitably used in combination with an association study-prior odds are higher for candidate genes mapping into a QTL region in the QTL mapping family, and, optionally, the number of markers genotyped in an association study can be reduced. Examples using R and the R packages BayesQTLBIC, ncdf are given.

Designing a GWAS: power, sample size, and data structure

In this chapter we describe a novel Bayesian approach to designing GWAS studies with the goal of ensuring robust detection of effects of genomic loci associated with trait variation.The goal of GWAS is to detect loci associated with variation in traits of interest. Finding which of 500,000-1,000,000 loci has a practically significant effect is a difficult statistical problem, like finding a needle in a haystack. We address this problem by designing experiments to detect effects with a given Bayes factor, where the Bayes factor is chosen sufficiently large to overcome the low prior odds for genomic associations. Methods are given for various possible data structures including random population samples, case-control designs, transmission disequilibrium tests, sib-based transmission disequilibrium tests, and other family-based designs including designs for plants with clonal replication. We also consider the problem of eliciting prior information from experts, which is necessary to quantify prior odds for loci. We advocate a "subjective" Bayesian approach, where the prior distribution is considered as a mathematical representation of our prior knowledge, while also giving generic formulae that allow conservative computations based on low prior information, e.g., equivalent to the information in a single sample point. Examples using R and the R packages ldDesign are given throughout.

Meta-analysis of QTL mapping experiments

Meta-analysis is an important tool for integrating information from multiple quantitative trait loci (QTLs) studies. Pooling of results from several studies allows greater statistical power for QTL detection and more precise estimation of their genetic effects. Hence, a meta-analysis can yield conclusions that are stronger than those of individual studies and can give greater insight into the genetic architecture of complex traits. In this chapter, we present basic theories and methods for meta-analysis of QTL mapping experiments. The meta-analytic procedures are described in a general context. The statistical methods cover both parametric and nonparametric statistical models. Finally, we illustrate the features of these statistical methods using simulated and real datasets.

Experimental designs for robust detection of effects in genome-wide case-control studies

In genome-wide association studies hundreds of thousands of loci are scanned in thousands of cases and controls, with the goal of identifying genomic loci underpinning disease. This is a challenging statistical problem requiring strong evidence. Only a small proportion of the heritability of common diseases has so far been explained. This "dark matter of the genome" is a subject of much discussion. It is critical to have experimental design criteria that ensure that associations between genomic loci and phenotypes are robustly detected. To ensure associations are robustly detected we require good power (e.g., 0.8) and sufficiently strong evidence [i.e., a high Bayes factor (e.g., 10(6), meaning the data are 1 million times more likely if the association is real than if there is no association)] to overcome the low prior odds for any given marker in a genome scan to be associated with a causal locus. Power calculations are given for determining the sample sizes necessary to detect effects with the required power and Bayes factor for biallelic markers in linkage disequilibrium with causal loci in additive, dominant, and recessive genetic models. Significantly stronger evidence and larger sample sizes are required than indicated by traditional hypothesis tests and power calculations. Many reported putative effects are not robustly detected and many effects including some large moderately low-frequency effects may remain undetected. These results may explain the dark matter in the genome. The power calculations have been implemented in R and will be available in the R package ldDesign.

Genome-wide association studies for osteochondrosis in French Trotter horses

A genome-wide association study for osteochondrosis (OC) in French Trotter horses was carried out to detect QTL using genotype data from the Illumina EquineSNP50 BeadChip assay. Analysis data came from 161 sire families of French Trotter horses with 525 progeny and family sizes ranging from 1 to 20. Genotypes were available for progeny (n = 525) and sires with at least 2 progeny (n = 98). Radiographic data were obtained from progeny using at least 10 views to reveal OC. All radiographic findings were described by at least 2 veterinary experts in equine orthopedics, and severity indices (scores) were assigned based on the size and location of the lesion. Traits used were a global score, the sum of all severity scores lesions (GM, quantitative measurement), and the presence or absence of OC on the fetlock (FM), hock (HM), and other sites (other). Data were analyzed using 2 mixed models including fixed effects, polygenic effects, and SNP or haplotype cluster effects. By combining results with both methods at moderate evidence of association threshold P < 5 × 10(-5), this genome-wide association study displayed 1 region for GM on the Equus caballus chromosome (ECA) 13, 2 for HM on ECA 3 and 14, and 1 for other on ECA 15. One region on ECA 3 for HM represented the most significant hit (P = 3 × 10(-6)). By comparing QTL between traits at a decreased threshold (P < 5 × 10(-4)), the 4 QTL detected for GM were associated to a QTL detected for FM or HM but never both. Another interesting result was that no QTL were found in common between HM and FM.

Optimal use of regression models in genome-wide association studies

The performance of linear regression models in genome-wide association studies is influenced by how marker information is parameterized in the model. Considering the impact of parameterization is especially important when using information from multiple markers to test for association. Properties of the population, such as linkage disequilibrium (LD) and allele frequencies, will also affect the ability of a model to provide statistical support for an underlying quantitative trait locus (QTL). Thus, for a given location in the genome, the relationship between population properties and model parameterization is expected to influence the performance of the model in providing evidence for the position of a QTL. As LD and allele frequencies vary throughout the genome and between populations, understanding the relationship between these properties and model parameterization is of considerable importance in order to make optimal use of available genomic data. Here, we evaluate the performance of regression-based association models using genotype and haplotype information across the full spectrum of allele frequency and LD scenarios. Genetic marker data from 200 broiler chickens were used to simulate genomic conditions by selecting individual markers to act as surrogate QTL (sQTL) and then investigating the ability of surrounding markers to estimate sQTL genotypes and provide statistical support for their location. The LD and allele frequencies of markers and sQTL are shown to have a strong effect on the performance of models relative to one another. Our results provide an indication of the best choice of model parameterization given certain scenarios of marker and QTL LD and allele frequencies. We demonstrate a clear advantage of haplotype-based models, which account for phase uncertainty over other models tested, particularly for QTL with low minor allele frequencies. We show that the greatest advantage of haplotype models over single-marker models occurs when LD between markers and the causal locus is low. Under these situations, haplotype models have a greater accuracy of predicting the location of the QTL than other models tested.

Genome-wide association mapping in Norwegian Red cattle identifies quantitative trait loci for fertility and milk production on BTA12

Reproductive performance is a critical trait in dairy cattle. Poor reproductive performance leads to prolonged calving intervals, higher culling rates and extra expenses related to multiple inseminations, veterinary treatments and replacements. Genetic gain for improved reproduction through traditional selection is often slow because of low heritability and negative correlations with production traits. Detection of DNA markers associated with improved reproductive performance through genome-wide association studies could lead to genetic gain that is more balanced between fertility and production. Norwegian Red cattle are well suited for such studies, as very large numbers of detailed reproduction records are available. We conducted a genome-wide association study for non-return rate, fertility treatments and retained placenta using almost 1 million records on these traits and 17 343 genome-wide single-nucleotide polymorphisms. Genotyping costs were minimized by genotyping the sires of the cows recorded and by using daughter averages as phenotypes. The genotyped sires were assigned to either a discovery or a validation population. Associations were only considered to be validated if they were significant in both groups. Strong associations were found and validated on chromosomes 1, 5, 8, 9, 11 and 12. Several of these were highly supported by findings in other studies. The most important result was an association for non-return rate in heifers in a region of BTA12 where several associations for milk production traits have previously been found. Subsequent fine-mapping verified the presence of a quantitative trait loci (QTL) having opposing effects on non-return rate and milk production at 18 Mb. The other reproduction QTL did not have pleiotropic effects on milk production, and these are therefore of considerable interest for use in marker-assisted selection.

Application of association mapping to understanding the genetic diversity of plant germplasm resources

Compared to the conventional linkage mapping, linkage disequilibrium (LD)-mapping, using the nonrandom associations of loci in haplotypes, is a powerful high-resolution mapping tool for complex quantitative traits. The recent advances in the development of unbiased association mapping approaches for plant population with their successful applications in dissecting a number of simple to complex traits in many crop species demonstrate a flourish of the approach as a “powerful gene tagging” tool for crops in the plant genomics era of 21st century. The goal of this review is to provide nonexpert readers of crop breeding community with (1) the basic concept, merits, and simple description of existing methodologies for an association mapping with the recent improvements for plant populations, and (2) the details of some of pioneer and recent studies on association mapping in various crop species to demonstrate the feasibility, success, problems, and future perspectives of the efforts in plants. This should be helpful for interested readers of international plant research community as a guideline for the basic understanding, choosing the appropriate methods, and its application.

Allelic variation in cell wall candidate genes affecting solid wood properties in natural populations and land races of Pinus radiata

Forest trees are ideally suited to association mapping due to their high levels of diversity and low genomic linkage disequilibrium. Using an association mapping approach, single-nucleotide polymorphism (SNP) markers influencing quantitative variation in wood quality were identified in a natural population of Pinus radiata. Of 149 sites examined, 10 demonstrated significant associations (P < 0.05, q < 0.1) with one or more traits after accounting for population structure and experimentwise error. Without accounting for marker interactions, phenotypic variation attributed to individual SNPs ranged from 2 to 6.5%. Undesirable negative correlations between wood quality and growth were not observed, indicating potential to break negative correlations by selecting for individual SNPs in breeding programs. Markers that yielded significant associations were reexamined in an Australian land race. SNPs from three genes (PAL1, PCBER, and SUSY) yielded significant associations. Importantly, associations with two of these genes validated associations with density previously observed in the discovery population. In both cases, decreased wood density was associated with the minor allele, suggesting that these SNPs may be under weak negative purifying selection for density in the natural populations. These results demonstrate the utility of LD mapping to detect associations, even when the power to detect SNPs with small effect is anticipated to be low.

A genome wide association study for QTL affecting direct and maternal effects of stillbirth and dystocia in cattle

Dystocia and stillbirth are significant causes of female and neonatal death in many species and there is evidence for a genetic component to both traits. Identifying causal mutations affecting these traits through genome wide association studies could reveal the genetic pathways involved and will be a step towards targeted interventions. Norwegian Red cattle are an ideal model breed for such studies as very large numbers of records are available. We conducted a genome wide association study for direct and maternal effects of dystocia and stillbirth using almost 1 million records of these traits. Genotyping costs were minimized by genotyping the sires of the recorded cows, and using daughter averages as phenotypes. A dense marker map containing 17,343 single nucleotide polymorphisms covering all autosomal chromosomes was utilized. The genotyped sires were assigned to one of two groups in an attempt to ensure independence between the groups. Associations were only considered validated if they occurred in both groups. Strong associations were found and validated on chromosomes 4, 5, 6, 9, 12, 20, 22 and 28. The QTL region on chromosome 6 was refined using LDLA analysis. The results showed that this chromosome most probably contains two QTL for direct effect on dystocia and one for direct effect on stillbirth. Several candidate genes may be identified close to these QTL. Of these, a cluster of genes expected to affect bone and cartilage formation (i.e. SPP1, IBSP and MEPE) are of particular interest and we suggest that these genes are screened in candidate gene studies for dystocia and stillbirth in cattle as well as other species.

Linkage disequilibrium based association mapping of fiber quality traits in G. hirsutum L. variety germplasm

Cotton is the world's leading cash crop, but it lags behind other major crops for marker-assisted breeding due to limited polymorphisms and a genetic bottleneck through historic domestication. This underlies a need for characterization, tagging, and utilization of existing natural polymorphisms in cotton germplasm collections. Here we report genetic diversity, population characteristics, the extent of linkage disequilibrium (LD), and association mapping of fiber quality traits using 202 microsatellite marker primer pairs in 335 G. hirsutum germplasm grown in two diverse environments, Uzbekistan and Mexico. At the significance threshold (r (2) >or= 0.1), a genome-wide average of LD extended up to genetic distance of 25 cM in assayed cotton variety accessions. Genome wide LD at r (2) >or= 0.2 was reduced to approximately 5-6 cM, providing evidence of the potential for association mapping of agronomically important traits in cotton. Results suggest linkage, selection, inbreeding, population stratification, and genetic drift as the potential LD-generating factors in cotton. In two environments, an average of ~20 SSR markers was associated with each main fiber quality traits using a unified mixed liner model (MLM) incorporating population structure and kinship. These MLM-derived significant associations were confirmed in general linear model and structured association test, accounting for population structure and permutation-based multiple testing. Several common markers, showing the significant associations in both Uzbekistan and Mexican environments, were determined. Between 7 and 43% of the MLM-derived significant associations were supported by a minimum Bayes factor at 'moderate to strong' and 'strong to very strong' evidence levels, suggesting their usefulness for marker-assisted breeding programs and overall effectiveness of association mapping using cotton germplasm resources.

Identification of associations between SSR markers and fiber traits in an exotic germplasm derived from multiple crosses among Gossypium tetraploid species

Genetic improvement in yield and fiber quality is needed for worldwide cotton production. Identification of molecular markers associated with fiber-related traits can facilitate selection for these traits in breeding. This study was designed to identify associations between SSR markers and fiber traits using an exotic germplasm population, species polycross (SP), derived from multiple crosses among Gossypium tetraploid species. The SP population underwent 11 generations of mixed random mating and selfing followed by 12 generations of selfing. A total of 260 lines were evaluated for fiber-related traits under three environments in 2005 and 2006. Large genotypic variance components in traits were identified relative to components of genotype x environment. Eighty-six primer pairs amplified a total of 314 polymorphic fragments among 260 lines. A total of 202 fragments with above 6% allele frequency were analyzed for associations. Fifty-nine markers were found to have a significant (P < 0.05, 0.01, or 0.001) association with six fiber traits. There were six groups identified within the population using structure analysis. Allele frequency divergence among six groups ranged from 0.11 to 0.27. Of the 59 marker-trait associations, 39 remained significant after correction for population structure and kinship using a mixed linear model. The effect of population sub-structure on associations was most significant in boll weight among the traits analyzed. The sub-structure among the SP lines may be caused by natural selection, the breeding method applied during development of inbred lines, and unknown factors. The identified marker-trait associations can be useful in breeding and help determine genetic mechanisms underlying interrelationships among fiber traits.

Genotype matrix mapping: searching for quantitative trait loci interactions in genetic variation in complex traits

In order to reveal quantitative trait loci (QTL) interactions and the relationship between various interactions in complex traits, we have developed a new QTL mapping approach, named genotype matrix mapping (GMM), which searches for QTL interactions in genetic variation. The central approach in GMM is the following. (1) Each tested marker is given a virtual matrix, named a genotype matrix (GM), containing intersecting lines and rows equal to the total allele number for that marker in the population analyzed. (2) QTL interactions are then estimated and compared through virtual networks among the GMs. To evaluate the contribution of marker combinations to a quantitative phenotype, the GMM method divides the samples into two non-overlapping subclasses, S₀ and S₁; the former contains the samples that have a specific genotype pattern to be evaluated, and the latter contains samples that do not. Based on this division, the F-measure is calculated as an index of significance. With the GMM method, we extracted significant marker combinations consisting of one to three interacting markers. The results indicated there were multiple QTL interactions affecting the phenotype (flowering date). GMM will be a valuable approach to identify QTL interactions in genetic variation of a complex trait within a variety of organisms.

Mapping determinants of human gene expression by regional and genome-wide association

To study the genetic basis of natural variation in gene expression, we previously carried out genome-wide linkage analysis and mapped the determinants of approximately 1,000 expression phenotypes. In the present study, we carried out association analysis with dense sets of single-nucleotide polymorphism (SNP) markers from the International HapMap Project. For 374 phenotypes, the association study was performed with markers only from regions with strong linkage evidence; these regions all mapped close to the expressed gene. For a subset of 27 phenotypes, analysis of genome-wide association was performed with >770,000 markers. The association analysis with markers under the linkage peaks confirmed the linkage results and narrowed the candidate regulatory regions for many phenotypes with strong linkage evidence. The genome-wide association analysis yielded highly significant results that point to the same locations as the genome scans for about 50% of the phenotypes. For one candidate determinant, we carried out functional analyses and confirmed the variation in cis-acting regulatory activity. Our findings suggest that association studies with dense SNP maps will identify susceptibility loci or other determinants for some complex traits or diseases.