A new tool called DISSECT for analysing large genomic data sets using a Big Data approach

Investigating pedigree- and SNP-associated components of heritability in a wild population of Soay sheep

Estimates of narrow sense heritability derived from genomic data that contain related individuals may be biased due to the within-family effects such as dominance, epistasis and common environmental factors. However, for many wild populations, removal of related individuals from the data would result in small sample sizes. In 2013, Zaitlen et al. proposed a method to estimate heritability in populations that include close relatives by simultaneously fitting an identity-by-state (IBS) genomic relatedness matrix (GRM) and an identity-by-descent (IBD) GRM. The IBD GRM is identical to the IBS GRM, except relatedness estimates below a specified threshold are set to 0. We applied this method to a sample of 8557 wild Soay sheep from St. Kilda, with genotypic information for 419,281 single nucleotide polymorphisms. We aimed to see how this method would partition heritability into population-level (IBS) and family-associated (IBD) variance for a range of genetic architectures, and so we focused on a mixture of polygenic and monogenic traits. We also implemented a variant of the model in which the IBD GRM was replaced by a GRM constructed from SNPs with low minor allele frequency to examine whether any additive genetic variance is captured by rare alleles. Whilst the inclusion of the IBD GRM did not significantly improve the fit of the model for the monogenic traits, it improved the fit for some of the polygenic traits, suggesting that dominance, epistasis and/or common environment not already captured by the non-genetic random effects fitted in our models may influence these traits.

The impact of SNP density on quantitative genetic analyses of body size traits in a wild population of Soay sheep

Understanding the genetic architecture underpinning quantitative traits in wild populations is pivotal to understanding the processes behind trait evolution. The 'animal model' is a popular method for estimating quantitative genetic parameters such as heritability and genetic correlation and involves fitting an estimate of relatedness between individuals in the study population. Genotypes at genome-wide markers can be used to estimate relatedness; however, relatedness estimates vary with marker density, potentially affecting results. Increasing density of markers is also expected to increase the power to detect quantitative trait loci (QTL). In order to understand how the density of genetic markers affects the results of quantitative genetic analyses, we estimated heritability and performed genome-wide association studies (GWAS) on five body size traits in an unmanaged population of Soay sheep using two different SNP densities: a dataset of 37,037 genotyped SNPs and an imputed dataset of 417,373 SNPs. Heritability estimates did not differ between the two SNP densities, but the high-density imputed SNP dataset revealed four new SNP-trait associations that were not found with the lower density dataset, as well as confirming all previously-found QTL. We also demonstrated that fitting fixed and random effects in the same step as performing GWAS is a more powerful approach than pre-correcting for covariates in a separate model.

A generalized linear mixed model association tool for biobank-scale data

Compared with linear mixed model-based genome-wide association (GWA) methods, generalized linear mixed model (GLMM)-based methods have better statistical properties when applied to binary traits but are computationally much slower. In the present study, leveraging efficient sparse matrix-based algorithms, we developed a GLMM-based GWA tool, fastGWA-GLMM, that is severalfold to orders of magnitude faster than the state-of-the-art tools when applied to the UK Biobank (UKB) data and scalable to cohorts with millions of individuals. We show by simulation that the fastGWA-GLMM test statistics of both common and rare variants are well calibrated under the null, even for traits with extreme case-control ratios. We applied fastGWA-GLMM to the UKB data of 456,348 individuals, 11,842,647 variants and 2,989 binary traits (full summary statistics available at http://fastgwa.info/ukbimpbin ), and identified 259 rare variants associated with 75 traits, demonstrating the use of imputed genotype data in a large cohort to discover rare variants for binary complex traits.

Sex differences in genetic architecture in the UK Biobank

Males and females present differences in complex traits and in the risk of a wide array of diseases. Genotype by sex (GxS) interactions are thought to account for some of these differences. However, the extent and basis of GxS are poorly understood. In the present study, we provide insights into both the scope and the mechanism of GxS across the genome of about 450,000 individuals of European ancestry and 530 complex traits in the UK Biobank. We found small yet widespread differences in genetic architecture across traits. We also found that, in some cases, sex-agnostic analyses may be missing trait-associated loci and looked into possible improvements in the prediction of high-level phenotypes. Finally, we studied the potential functional role of the differences observed through sex-biased gene expression and gene-level analyses. Our results suggest the need to consider sex-aware analyses for future studies to shed light onto possible sex-specific molecular mechanisms.

Epigenetic prediction of major depressive disorder

Variation in DNA methylation (DNAm) is associated with lifestyle factors such as smoking and body mass index (BMI) but there has been little research exploring its ability to identify individuals with major depressive disorder (MDD). Using penalised regression on genome-wide CpG methylation, we tested whether DNAm risk scores (MRS), trained on 1223 MDD cases and 1824 controls, could discriminate between cases (n = 363) and controls (n = 1417) in an independent sample, comparing their predictive accuracy to polygenic risk scores (PRS). The MRS explained 1.75% of the variance in MDD (β = 0.338, p = 1.17 × 10-7) and remained associated after adjustment for lifestyle factors (β = 0.219, p = 0.001, R2 = 0.68%). When modelled alongside PRS (β = 0.384, p = 4.69 × 10-9) the MRS remained associated with MDD (β = 0.327, p = 5.66 × 10-7). The MRS was also associated with incident cases of MDD who were well at recruitment but went on to develop MDD at a later assessment (β = 0.193, p = 0.016, R2 = 0.52%). Heritability analyses found additive genetic effects explained 22% of variance in the MRS, with a further 19% explained by pedigree-associated genetic effects and 16% by the shared couple environment. Smoking status was also strongly associated with MRS (β = 0.440, p ≤ 2 × 10-16). After removing smokers from the training set, the MRS strongly associated with BMI (β = 0.053, p = 0.021). We tested the association of MRS with 61 behavioural phenotypes and found that whilst PRS were associated with psychosocial and mental health phenotypes, MRS were more strongly associated with lifestyle and sociodemographic factors. DNAm-based risk scores of MDD significantly discriminated MDD cases from controls in an independent dataset and may represent an archive of exposures to lifestyle factors that are relevant to the prediction of MDD.

Regional heritability mapping identifies several novel loci (STAT4, ULK4, and KCNH5) for primary biliary cholangitis in the Japanese population

While the advent of GWAS more than a decade ago has ushered in remarkable advances in our understanding of complex traits, the limitations of single-SNP analysis have also led to the development of several other approaches. Simulation studies have shown that the regional heritability mapping (RHM) method, which makes use of multiple adjacent SNPs jointly to estimate the genetic effect of a given region of the genome, generally has higher detection power than single-SNP GWAS. However, thus far its use has been mostly limited to agricultural settings, and its potential for the discovery of new genes in human diseases is yet to be fully exploited. In this study, by applying the RHM method to primary biliary cholangitis (PBC) in the Japanese population, we identified three novel loci (STAT4, ULK4, and KCNH5) at the genome-wide significance level, two of which (ULK4 and KCNH5) have not been found associated with PBC in any population previously. Notably, these genes could not be detected by using conventional single-SNP GWAS, highlighting the potential of the RHM method for the detection of new susceptibility loci in human diseases. These findings thereby provide strong empirical evidence that RHM is an effective and practical complementary approach to GWAS in this context. Also, liver tissue mRNA microarray analysis revealed higher gene expression levels in ULK4 in PBC patients (P < 0.01). Lastly, we estimated the common SNP heritability of PBC in the Japanese population (0.210 ± 0.026).

Patterns of MHC-dependent sexual selection in a free-living population of sheep

The MHC is one of the most polymorphic gene clusters in vertebrates and play an essential role in adaptive immunity. Apart from pathogen-mediated selection, sexual selection can also contribute to the maintenance of MHC diversity. MHC-dependent sexual selection could occur via several mechanisms but at present there is no consensus as to which of these mechanisms are involved and their importance. Previous studies have often suffered from limited genetic and behavioural data and small sample size, and were rarely able to examine all the mechanisms together, determine whether signatures of MHC-based non-random mating are independent of genomic effects or differentiate whether MHC-dependent sexual selection takes place at the pre- or post-copulatory stage. In this study, we use Monte Carlo simulation to investigate evidence for non-random MHC-dependent mating patterns by all three mechanisms in a free-living population of Soay sheep. Using 1710 sheep diplotyped at the MHC class IIa region and genome-wide SNPs, together with field observations of consorts, we found sexual selection against a particular haplotype in males at the pre-copulatory stage and sexual selection against female MHC heterozygosity during the rut. We also found MHC-dependent disassortative mating at the post-copulatory stage, along with strong evidence of inbreeding avoidance at both stages. However, results from generalized linear mixed models suggest that the pattern of MHC-dependent disassortative mating could be a by-product of inbreeding avoidance. Our results therefore suggest that while multiple apparent mechanisms of non-random mating with respect to the MHC may occur, some of them have alternative explanations.

Identification of Major Loci and Candidate Genes for Meat Production-Related Traits in Broilers

Background

Carcass traits are crucial characteristics of broilers. However, the underlying genetic mechanisms are not well understood. In the current study, significant loci and major-effect candidate genes affecting nine carcass traits related to meat production were analyzed in 873 purebred broilers using an imputation-based genome-wide association study.

Results

The heritability estimates of nine carcass traits, including carcass weight, thigh muscle weight, and thigh muscle percentage, were moderate to high and ranged from 0.21 to 0.39. Twelve genome-wide significant SNPs and 118 suggestively significant SNPs of 546,656 autosomal variants were associated with carcass traits. All SNPs for six weight traits (body weight at 42 days of age, carcass weight, eviscerated weight, whole thigh weight, thigh weight, and thigh muscle weight) were clustered around the 24.08 Kb region (GGA24: 5.73–5.75 Mb) and contained only one candidate gene (DRD2). The most significant SNP, rs15226023, accounted for 4.85–7.71% of the estimated genetic variance of the six weight traits. The remaining SNPs for carcass composition traits (whole thigh percentage and thigh percentage) were clustered around the 42.52 Kb region (GGA3: 53.03–53.08 Mb) and contained only one candidate gene (ADGRG6). The most significant SNP in this region, rs13571431, accounted for 11.89–13.56% of the estimated genetic variance of two carcass composition traits. Some degree of genetic differentiation in ADGRG6 between large and small breeds was observed.

Conclusion

We identified one 24.08 Kb region for weight traits and one 42.52 Kb region for thigh-related carcass traits. DRD2 was the major-effect candidate gene for weight traits, and ADGRG6 was the major-effect candidate gene for carcass composition traits. Our results supply essential information for causative mutation identification of carcass traits in broilers.

Evidence of horizontal indirect genetic effects in humans

Indirect genetic effects, the effects of the genotype of one individual on the phenotype of other individuals, are environmental factors associated with human disease and complex trait variation that could help to expand our understanding of the environment linked to complex traits. Here, we study indirect genetic effects in 80,889 human couples of European ancestry for 105 complex traits. Using a linear mixed model approach, we estimate partner indirect heritability and find evidence of partner heritability on ~50% of the analysed traits. Follow-up analysis suggests that in at least ~25% of these traits, the partner heritability is consistent with the existence of indirect genetic effects including a wide variety of traits such as dietary traits, mental health and disease. This shows that the environment linked to complex traits is partially explained by the genotype of other individuals and motivates the need to find new ways of studying the environment.

Quantitative trait loci and transcriptome signatures associated with avian heritable resistance to Campylobacter

Campylobacter is the leading cause of bacterial foodborne gastroenteritis worldwide. Handling or consumption of contaminated poultry meat is a key risk factor for human campylobacteriosis. One potential control strategy is to select poultry with increased resistance to Campylobacter. We associated high-density genome-wide genotypes (600K single nucleotide polymorphisms) of 3000 commercial broilers with Campylobacter load in their caeca. Trait heritability was modest but significant (h² = 0.11 ± 0.03). Results confirmed quantitative trait loci (QTL) on chromosomes 14 and 16 previously identified in inbred chicken lines, and detected two additional QTLs on chromosomes 19 and 26. RNA-Seq analysis of broilers at the extremes of colonisation phenotype identified differentially transcribed genes within the QTL on chromosome 16 and proximal to the major histocompatibility complex (MHC) locus. We identified strong cis-QTLs located within MHC suggesting the presence of cis-acting variation in MHC class I and II and BG genes. Pathway and network analyses implicated cooperative functional pathways and networks in colonisation, including those related to antigen presentation, innate and adaptive immune responses, calcium, and renin–angiotensin signalling. While co-selection for enhanced resistance and other breeding goals is feasible, the frequency of resistance-associated alleles was high in the population studied and non-genetic factors significantly influenced Campylobacter colonisation.

Genome-Wide Association Study of Piscine Myocarditis Virus (PMCV) Resistance in Atlantic Salmon (Salmo salar)

Cardiomyopathy syndrome is a severe, viral disease of Atlantic salmon that mostly affects farmed animals during their late production stage at sea. Caused by piscine myocarditis virus (PMCV), over the past few years outbreaks due to this disease have resulted in significant losses to the aquaculture industry. However, there is currently no vaccine that has proven effective against this virus. In this study, using a challenge model, we investigated the genetic variation for resistance to PMCV, by screening a large number of animals using a 55 K SNP array. In particular, we aimed to identify genetic markers that are tightly linked to higher disease resistance and can potentially be used in breeding programs. Using genomic information, we estimated a heritability of 0.51 ± 0.06, suggesting that resistance against this virus, to a great extent, is controlled by genetic factors. Through association analysis, we identified a significant quantitative trait locus (QTL) on chromosome 27, explaining approximately 57% of the total additive genetic variation. The region harboring this QTL contains various immune-related candidate genes, many of which have previously been shown to have a different expression profile between the naïve and infected animals. We also identified a suggestive association on chromosome 12, with the QTL linked markers located in 2 putatively immune-related genes. These results are of particular interest, as they can readily be implemented into breeding programs, can further assist in fine-mapping the causative mutations, and help in better understanding the biology of the disease and the immunological mechanisms underlying resistance against PMCV.

Identification of genetic loci associated with higher resistance to pancreas disease (PD) in Atlantic salmon (Salmo salar L.)

Background

Pancreas disease (PD) is a contagious disease caused by salmonid alphavirus (SAV) with significant economic and welfare impacts on salmon farming. Previous work has shown that higher resistance against PD has underlying additive genetic components and can potentially be improved through selective breeding. To better understand the genetic basis of PD resistance in Atlantic salmon, we challenged 4506 smolts from 296 families of the SalmoBreed strain. Fish were challenged through intraperitoneal injection with the most virulent form of the virus found in Norway (i.e., SAV3). Mortalities were recorded, and more than 900 fish were further genotyped on a 55 K SNP array.

Results

The estimated heritability for PD resistance was 0.41 ± 0.017. The genetic markers on two chromosomes, ssa03 and ssa07, showed significant associations with higher disease resistance. Collectively, markers on these two QTL regions explained about 60% of the additive genetic variance. We also sequenced and compared the cardiac transcriptomics of moribund fish and animals that survived the challenge with a focus on candidate genes within the chromosomal segments harbouring QTL. Approximately 200 genes, within the QTL regions, were found to be differentially expressed. Of particular interest, we identified various components of immunoglobulin-heavy-chain locus B (IGH-B) on ssa03 and immunoglobulin-light-chain on ssa07 with markedly higher levels of transcription in the resistant animals. These genes are closely linked to the most strongly QTL associated SNPs, making them likely candidates for further investigation.

Conclusions

The findings presented here provide supporting evidence that breeding is an efficient tool for increasing PD resistance in Atlantic salmon populations. The estimated heritability is one of the largest reported for any disease resistance in this species, where the majority of the genetic variation is explained by two major QTL. The transcriptomic analysis has revealed the activation of essential components of the innate and the adaptive immune responses following infection with SAV3. Furthermore, the complementation of the genomic with the transcriptomic data has highlighted the possible critical role of the immunoglobulin loci in combating PD virus.

Scalable probabilistic PCA for large-scale genetic variation data

Principal component analysis (PCA) is a key tool for understanding population structure and controlling for population stratification in genome-wide association studies (GWAS). With the advent of large-scale datasets of genetic variation, there is a need for methods that can compute principal components (PCs) with scalable computational and memory requirements. We present ProPCA, a highly scalable method based on a probabilistic generative model, which computes the top PCs on genetic variation data efficiently. We applied ProPCA to compute the top five PCs on genotype data from the UK Biobank, consisting of 488,363 individuals and 146,671 SNPs, in about thirty minutes. To illustrate the utility of computing PCs in large samples, we leveraged the population structure inferred by ProPCA within White British individuals in the UK Biobank to identify several novel genome-wide signals of recent putative selection including missense mutations in RPGRIP1L and TLR4.

Principal component analysis is a commonly used technique for understanding population structure and genetic variation. With the advent of large-scale datasets that contain the genetic information of hundreds of thousands of individuals, there is a need for methods that can compute principal components (PCs) with scalable computational and memory requirements. In this study, we present ProPCA, a highly scalable statistical method to compute genetic PCs efficiently. We systematically evaluate the accuracy and scalability of our method on large-scale simulated data and apply it to the UK Biobank. Leveraging the population structure inferred by ProPCA within the White British individuals in the UK Biobank, we identify several novel signals of putative recent selection.

Identification of genomic regions regulating sex determination in Atlantic salmon using high density SNP data

Background

A complete understanding of the genetic basis for sexual determination and differentiation is necessary in order to implement efficient breeding schemes at early stages of development. Atlantic salmon belongs to the family Salmonidae of fishes and represents a species of great commercial value. Although the species is assumed to be male heterogametic with XY sex determination, the precise genetic basis of sexual development remains unclear. The complexity is likely associated to the relatively recent salmonid specific whole genome duplication that may be responsible for certain genome instability. This instability together with the capacity of the sex-determining gene to move across the genome as reported by previous studies, may explain that sexual development genes are not circumscribed to the same chromosomes in all members of the species. In this study, we have used a 220 K SNP panel developed for Atlantic salmon to identify the chromosomes explaining the highest proportion of the genetic variance for sex as well as candidate regions and genes associated to sexual development in this species.

Results

Results from regional heritability analysis showed that the chromosomes explaining the highest proportion of variance in these populations were Ssa02 (heritability = 0.42, SE = 0.12) and Ssa21 (heritability = 0.26, SE = 0.11). After pruning by linkage disequilibrium, genome-wide association analyses revealed 114 SNPs that were significantly associated with sex, being Ssa02 the chromosome containing a greatest number of regions. Close examination of the candidate regions evidenced important genes related to sex in other species of Class Actinopterygii, including SDY, genes from family SOX, RSPO1, ESR1, U2AF2A, LMO7, GNRH-R, DND and FIGLA.

Conclusions

The combined results from regional heritability analysis and genome-wide association have provided new advances in the knowledge of the genetic regulation of sex determination in Atlantic salmon, supporting that Ssa02 is the candidate chromosome for sex in this species and suggesting an alternative population lineage in Spanish wild populations according to the results from Ssa21.

Genomic diversity, linkage disequilibrium and selection signatures in European local pig breeds assessed with a high density SNP chip

Genetic characterization of local breeds is essential to preserve their genomic variability, to advance conservation policies and to contribute to their promotion and sustainability. Genomic diversity of twenty European local pig breeds and a small sample of Spanish wild pigs was assessed using high density SNP chips. A total of 992 DNA samples were analyzed with the GeneSeek Genomic Profiler (GGP) 70 K HD porcine genotyping chip. Genotype data was employed to compute genetic diversity, population differentiation and structure, genetic distances, linkage disequilibrium and effective population size. Our results point out several breeds, such as Turopolje, Apulo Calabrese, Casertana, Mora Romagnola and Lithuanian indigenous wattle, having the lowest genetic diversity, supported by low heterozygosity and very small effective population size, demonstrating the need of enhanced conservation strategies. Principal components analysis showed the clustering of the individuals of the same breed, with few breeds being clearly isolated from the rest. Several breeds were partially overlapped, suggesting genetic closeness, which was particularly marked in the case of Iberian and Alentejana breeds. Spanish wild boar was also narrowly related to other western populations, in agreement with recurrent admixture between wild and domestic animals. We also searched across the genome for loci under diversifying selection based on F_ST outlier tests. Candidate genes that may underlie differences in adaptation to specific environments and productive systems and phenotypic traits were detected in potentially selected genomic regions.

Discovery and Functional Annotation of Quantitative Trait Loci Affecting Resistance to Sea Lice in Atlantic Salmon

Sea lice (Caligus rogercresseyi) are ectoparasitic copepods which have a large negative economic and welfare impact in Atlantic salmon (Salmo salar) aquaculture, particularly in Chile. A multi-faceted prevention and control strategy is required to tackle lice, and selective breeding contributes via cumulative improvement of host resistance to the parasite. While host resistance has been shown to be heritable, little is yet known about the individual loci that contribute to this resistance, the potential underlying genes, and their mechanisms of action. In this study we took a multifaceted approach to identify and characterize quantitative trait loci (QTL) affecting host resistance in a population of 2,688 Caligus-challenged Atlantic salmon post-smolts from a commercial breeding program. We used low and medium density genotyping with imputation to collect genome-wide SNP marker data for all animals. Moderate heritability estimates of 0.28 and 0.24 were obtained for lice density (as a measure of host resistance) and growth during infestation, respectively. Three QTL explaining between 7 and 13% of the genetic variation in resistance to sea lice (as represented by the traits of lice density) were detected on chromosomes 3, 18, and 21. Characterisation of these QTL regions was undertaken using RNA sequencing and pooled whole genome sequencing data. This resulted in the identification of a shortlist of potential underlying causative genes, and candidate functional mutations for further study. For example, candidates within the chromosome 3 QTL include a putative premature stop mutation in TOB1 (an anti-proliferative transcription factor involved in T cell regulation) and an uncharacterized protein which showed significant differential allelic expression (implying the existence of a cis-acting regulatory mutation). While host resistance to sea lice is polygenic in nature, the results of this study highlight significant QTL regions together explaining between 7 and 13 % of the heritability of the trait. Future investigation of these QTL may enable improved knowledge of the functional mechanisms of host resistance to sea lice, and incorporation of functional variants to improve genomic selection accuracy.

Indirect assortative mating for human disease and longevity

Phenotypic correlations among partners for traits such as longevity or late-onset disease have been found to be comparable to phenotypic correlations in first-degree relatives. How these correlations arise in late life is poorly understood. Here we introduce a novel paradigm to establish the presence of indirect assortment on factors correlated across generations, by examining correlations between parents of couples, i.e., in-laws. Using correlations in additive genetic values we further corroborate the presence of indirect assortment on heritable factors. Specifically, using couples from the UK Biobank cohort, we show that longevity and disease history of the parents of White British couples are correlated, with correlations of up to 0.09. The correlations in parental longevity are replicated in the FamiLinx cohort, a larger and geographically more diverse historical ancestry dataset spanning a broader time frame. These correlations in parental longevity significantly (pval < 0.0093 for all pairs of parents) exceed what would be expected due to variations in lifespan based on year and location of birth. For cardiovascular diseases, in particular hypertension, we find significant correlations (r = 0.028, pval = 0.005) in genetic values among partners, supporting a model where partners assort for risk factors to some extent genetically correlated with cardiovascular disease. Partitioning the relative importance of indirect assortative mating and shared common environment will require large, well-characterized longitudinal cohorts aimed at understanding phenotypic correlations among none-blood relatives. Identifying the factors that mediate indirect assortment on longevity and human disease risk will help to unravel factors affecting human disease and ultimately longevity.

Disentangling Genetic Variation for Resistance and Endurance to Scuticociliatosis in Turbot Using Pedigree and Genomic Information

Selective breeding for improving host responses to infectious pathogens is a promising option for disease control. In fact, disease resilience, the ability of a host to survive or cope with infectious challenge, has become a highly desirable breeding goal. However, resilience is a complex trait composed of two different host defence mechanisms, namely resistance (the ability of a host to avoid becoming infected or diseased) and endurance (the ability of an infected host to survive the infection). While both could be targeted for genetic improvement, it is currently unknown how they contribute to survival, as reliable estimates of genetic parameters for both traits obtained simultaneously are scarce. A difficulty lies in obtaining endurance phenotypes for genetic analyses. In this study, we present the results from an innovative challenge test carried out in turbot whose design allowed disentangling the genetic basis of resistance and endurance to Philasterides dicentrarchi, a parasite causing scuticociliatosis that leads to substantial economic losses in the aquaculture industry. A noticeable characteristic of the parasite is that it causes visual signs that can be used for disentangling resistance and endurance. Our results showed the existence of genetic variation for both traits (heritability = 0.26 and 0.12 for resistance and endurance, respectively) and for the composite trait resilience (heritability = 0.15). The genetic correlation between resistance and resilience was very high (0.90) indicating that both are at a large extent the same trait, but no significant genetic correlation was found between resistance and endurance. A total of 18,125 SNPs obtained from 2b-RAD sequencing enabled genome-wide association analyses for detecting QTLs controlling the three traits. A candidate QTL region on linkage group 19 that explains 33% of the additive genetic variance was identified for resilience. The region contains relevant genes related to immune response and defence mechanisms. Although no significant associations were found for resistance, the pattern of association was the same as for resilience. For endurance, one significant association was found on linkage group 2. The accuracy of genomic breeding values was also explored for resilience, showing that it increased by 12% when compared with the accuracy of pedigree-based breeding values. To our knowledge, this is the first study in turbot disentangling the genetic basis of resistance and endurance to scuticociliatosis.

Genetic architecture of human thinness compared to severe obesity

The variation in weight within a shared environment is largely attributable to genetic factors. Whilst many genes/loci confer susceptibility to obesity, little is known about the genetic architecture of healthy thinness. Here, we characterise the heritability of thinness which we found was comparable to that of severe obesity (h2 = 28.07 vs 32.33% respectively), although with incomplete genetic overlap (r = -0.49, 95% CI [-0.17, -0.82], p = 0.003). In a genome-wide association analysis of thinness (n = 1,471) vs severe obesity (n = 1,456), we identified 10 loci previously associated with obesity, and demonstrate enrichment for established BMI-associated loci (pbinomial = 3.05x10-5). Simulation analyses showed that different association results between the extremes were likely in agreement with additive effects across the BMI distribution, suggesting different effects on thinness and obesity could be due to their different degrees of extremeness. In further analyses, we detected a novel obesity and BMI-associated locus at PKHD1 (rs2784243, obese vs. thin p = 5.99x10-6, obese vs. controls p = 2.13x10-6 pBMI = 2.3x10-13), associations at loci recently discovered with much larger sample sizes (e.g. FAM150B and PRDM6-CEP120), and novel variants driving associations at previously established signals (e.g. rs205262 at the SNRPC/C6orf106 locus and rs112446794 at the PRDM6-CEP120 locus). Our ability to replicate loci found with much larger sample sizes demonstrates the value of clinical extremes and suggest that characterisation of the genetics of thinness may provide a more nuanced understanding of the genetic architecture of body weight regulation and may inform the identification of potential anti-obesity targets.

An atlas of genetic associations in UK Biobank

Genome-wide association studies (GWAS) have identified many loci contributing to variation in complex traits, yet the majority of loci that contribute to the heritability of complex traits remain elusive. Large study populations with sufficient statistical power are required to detect the small effect sizes of the yet unidentified genetic variants. However, the analysis of huge cohorts, like UK Biobank, is challenging. Here, we present an atlas of genetic associations for 118 non-binary and 660 binary traits of 452,264 UK Biobank participants of European ancestry. Results are compiled in a publicly accessible database that allows querying genome-wide association results for 9,113,133 genetic variants, as well as downloading GWAS summary statistics for over 30 million imputed genetic variants (>23 billion phenotype-genotype pairs). Our atlas of associations (GeneATLAS, http://geneatlas.roslin.ed.ac.uk ) will help researchers to query UK Biobank results in an easy and uniform way without the need to incur high computational costs.

A greedy feature selection algorithm for Big Data of high dimensionality

We present the Parallel, Forward-Backward with Pruning (PFBP) algorithm for feature selection (FS) for Big Data of high dimensionality. PFBP partitions the data matrix both in terms of rows as well as columns. By employing the concepts of p-values of conditional independence tests and meta-analysis techniques, PFBP relies only on computations local to a partition while minimizing communication costs, thus massively parallelizing computations. Similar techniques for combining local computations are also employed to create the final predictive model. PFBP employs asymptotically sound heuristics to make early, approximate decisions, such as Early Dropping of features from consideration in subsequent iterations, Early Stopping of consideration of features within the same iteration, or Early Return of the winner in each iteration. PFBP provides asymptotic guarantees of optimality for data distributions faithfully representable by a causal network (Bayesian network or maximal ancestral graph). Empirical analysis confirms a super-linear speedup of the algorithm with increasing sample size, linear scalability with respect to the number of features and processing cores. An extensive comparative evaluation also demonstrates the effectiveness of PFBP against other algorithms in its class. The heuristics presented are general and could potentially be employed to other greedy-type of FS algorithms. An application on simulated Single Nucleotide Polymorphism (SNP) data with 500K samples is provided as a use case.

Genome-wide association study in Japanese females identifies fifteen novel skin-related trait associations

Skin trait variation impacts quality-of-life, especially for females from the viewpoint of beauty. To investigate genetic variation related to these traits, we conducted a GWAS of various skin phenotypes in 11,311 Japanese women and identified associations for age-spots, freckles, double eyelids, straight/curly hair, eyebrow thickness, hairiness, and sweating. In silico annotation with RoadMap Epigenomics epigenetic state maps and colocalization analysis of GWAS and GTEx Project eQTL signals provided information about tissue specificity, candidate causal variants, and functional target genes. Novel signals for skin-spot traits neighboured AKAP1/MSI2 (rs17833789; P = 2.2 × 10-9), BNC2 (rs10810635; P = 2.1 × 10-22), HSPA12A (rs12259842; P = 7.1 × 10-11), PPARGC1B (rs251468; P = 1.3 × 10-21), and RAB11FIP2 (rs10444039; P = 5.6 × 10-21). HSPA12A SNPs were the only protein-coding gene eQTLs identified across skin-spot loci. Double edged eyelid analysis identified that a signal around EMX2 (rs12570134; P = 8.2 × 10-15) was also associated with expression of EMX2 and the antisense-RNA gene EMX2OS in brain putamen basal ganglia tissue. A known hair morphology signal in EDAR was associated with both eyebrow thickness (rs3827760; P = 1.7 × 10-9) and straight/curly hair (rs260643; P = 1.6 × 10-103). Excessive hairiness signals' top SNPs were also eQTLs for TBX15 (rs984225; P = 1.6 × 10-8), BCL2 (rs7226979; P = 7.3 × 10-11), and GCC2 and LIMS1 (rs6542772; P = 2.2 × 10-9). For excessive sweating, top variants in two signals in chr2:28.82-29.05 Mb (rs56089836; P = 1.7 × 10-11) were eQTLs for either PPP1CB or PLB1, while a top chr16:48.26-48.45 Mb locus SNP was a known ABCC11 missense variant (rs6500380; P = 6.8 × 10-10). In total, we identified twelve loci containing sixteen association signals, of which fifteen were novel. These findings will help dermatologic researchers better understand the genetic underpinnings of skin-related phenotypic variation in human populations.

Genome-Wide Association and Genomic Selection for Resistance to Amoebic Gill Disease in Atlantic Salmon

Amoebic gill disease (AGD) is one of the largest threats to salmon aquaculture, causing serious economic and animal welfare burden. Treatments can be expensive and environmentally damaging, hence the need for alternative strategies. Breeding for disease resistance can contribute to prevention and control of AGD, providing long-term cumulative benefits in selected stocks. The use of genomic selection can expedite selection for disease resistance due to improved accuracy compared to pedigree-based approaches. The aim of this work was to quantify and characterize genetic variation in AGD resistance in salmon, the genetic architecture of the trait, and the potential of genomic selection to contribute to disease control. An AGD challenge was performed in ∼1,500 Atlantic salmon, using gill damage and amoebic load as indicator traits for host resistance. Both traits are heritable (h2 ∼0.25-0.30) and show high positive correlation, indicating they may be good measurements of host resistance to AGD. While the genetic architecture of resistance appeared to be largely polygenic in nature, two regions on chromosome 18 showed suggestive association with both AGD resistance traits. Using a cross-validation approach, genomic prediction accuracy was up to 18% higher than that obtained using pedigree, and a reduction in marker density to ∼2,000 SNPs was sufficient to obtain accuracies similar to those obtained using the whole dataset. This study indicates that resistance to AGD is a suitable trait for genomic selection, and the addition of this trait to Atlantic salmon breeding programs can lead to more resistant stocks.

The Genomic Architecture of Fowl Typhoid Resistance in Commercial Layers

Salmonella enterica serovar Gallinarum causes devastating outbreaks of fowl typhoid across the globe, especially in developing countries. With the use of antimicrobial agents being reduced due to legislation and the absence of licensed vaccines in some parts of the world, an attractive complementary control strategy is to breed chickens for increased resistance to Salmonella. The potential for genetic control of salmonellosis has been demonstrated by experimental challenge of inbred populations. Quantitative trait loci (QTL) associated with resistance have been identified in many genomic regions. A major QTL associated with systemic salmonellosis has been identified in a region termed SAL1. In the present study, two outbreaks of fowl typhoid in 2007 and 2012 in the United Kingdom were used to investigate the genetic architecture of Salmonella resistance in commercial laying hens. In the first outbreak 100 resistant and 150 susceptible layers were genotyped using 11 single nucleotide polymorphism (SNP) and 3 microsatellite markers located in the previously identified SAL1 region on chromosome 5. From the second outbreak 100 resistant and 200 susceptible layers, belonging to a different line, were genotyped with a high-density (600 K) genome-wide SNP array. Substantial heritability estimates were obtained in both populations (h ² = 0.22 and 0.26, for the layers in the first and second outbreak, respectively). Significant associations with three markers on chromosome 5 located close to AKT1 and SIVA1 genes, coding for RAC-alpha serine/threonine protein kinase, and the CD27-binding protein SIVA1, respectively, were identified in the first outbreak. From analysis of the second outbreak, eight genome-wide significant associations with Salmonella resistance were identified on chromosomes 1, 6, 7, 11, 23, 24, 26, 28 and several others with suggestive genome-wide significance were found. Pathway and network analysis revealed the presence of many innate immune pathways related to Salmonella resistance. Although, significant associations with SNPs located in the SAL1 locus were not identified by the genome-wide scan for layers from the second outbreak, pathway analysis revealed P13K/AKT signaling as the most significant pathway. In summary, resistance to fowl typhoid is a heritable polygenic trait that could possibly be enhanced through selective breeding.

Leveraging Multi-ethnic Evidence for Risk Assessment of Quantitative Traits in Minority Populations

An essential component of precision medicine is the ability to predict an individual's risk of disease based on genetic and non-genetic factors. For complex traits and diseases, assessing the risk due to genetic factors is challenging because it requires knowledge of both the identity of variants that influence the trait and their corresponding allelic effects. Although the set of risk variants and their allelic effects may vary between populations, a large proportion of these variants were identified based on studies in populations of European descent. Heterogeneity in genetic architecture underlying complex traits and diseases, while broadly acknowledged, remains poorly characterized. Ignoring such heterogeneity likely reduces predictive accuracy for minority individuals. In this study, we propose an approach, called XP-BLUP, which ameliorates this ethnic disparity by combining trans-ethnic and ethnic-specific information. We build a polygenic model for complex traits that distinguishes candidate trait-relevant variants from the rest of the genome. The set of candidate variants are selected based on studies in any human population, yet the allelic effects are evaluated in a population-specific fashion. Simulation studies and real data analyses demonstrate that XP-BLUP adaptively utilizes trans-ethnic information and can substantially improve predictive accuracy in minority populations. At the same time, our study highlights the importance of the continued expansion of minority cohorts.

Fine-mapping host genetic variation underlying outcomes to Mycobacterium bovis infection in dairy cows

BACKGROUND

Susceptibility to Mycobacterium bovis infection in cattle is governed in part by host genetics. However, cattle diagnosed as infected with M. bovis display varying signs of pathology. The variation in host response to infection could represent a continuum since time of exposure or distinct outcomes due to differing pathogen handling. The relationships between host genetics and variation in host response and pathological sequelae following M. bovis infection were explored by genotyping 1966 Holstein-Friesian dairy cows at 538,231 SNPs with three distinct phenotypes. These were: single intradermal cervical comparative tuberculin (SICCT) test positives with visible lesions (VLs), SICCT-positives with undetected visible lesions (NVLs) and matched controls SICCT-negative on multiple occasions.

RESULTS

Regional heritability mapping identified three loci associated with the NVL phenotype on chromosomes 17, 22 and 23, distinct to the region on chromosome 13 associated with the VL phenotype. The region on chromosome 23 was at genome-wide significance and candidate genes overlapping the mapped window included members of the bovine leukocyte antigen class IIb region, a complex known for its role in immunity and disease resistance. Chromosome heritability analysis attributed variance to six and thirteen chromosomes for the VL and NVL phenotypes, respectively, and four of these chromosomes were found to explain a proportion of the phenotypic variation for both the VL and NVL phenotype. By grouping the M. bovis outcomes (VLs and NVLs) variance was attributed to nine chromosomes. When contrasting the two M. bovis infection outcomes (VLs vs NVLs) nine chromosomes were found to harbour heritable variation. Regardless of the case phenotype under investigation, chromosome heritability did not exceed 8% indicating that the genetic control of bTB resistance consists of variants of small to moderate effect situated across many chromosomes of the bovine genome.

CONCLUSIONS

These findings suggest the host genetics of M. bovis infection outcomes is governed by distinct and overlapping genetic variants. Thus, variation in the pathology of M. bovis infected cattle may be partly genetically determined and indicative of different host responses or pathogen handling. There may be at least three distinct outcomes following M. bovis exposure in dairy cattle: resistance to infection, infection resulting in pathology or no detectable pathology.

Multiple Trait Covariance Association Test Identifies Gene Ontology Categories Associated with Chill Coma Recovery Time in Drosophila melanogaster

The genomic best linear unbiased prediction (GBLUP) model has proven to be useful for prediction of complex traits as well as estimation of population genetic parameters. Improved inference and prediction accuracy of GBLUP may be achieved by identifying genomic regions enriched for causal genetic variants. We aimed at searching for patterns in GBLUP-derived single-marker statistics, by including them in genetic marker set tests, that could reveal associations between a set of genetic markers (genomic feature) and a complex trait. GBLUP-derived set tests proved to be powerful for detecting genomic features, here defined by gene ontology (GO) terms, enriched for causal variants affecting a quantitative trait in a population with low degree of relatedness. Different set test approaches were compared using simulated data illustrating the impact of trait- and genomic feature-specific factors on detection power. We extended the most powerful single trait set test, covariance association test (CVAT), to a multiple trait setting. The multiple trait CVAT (MT-CVAT) identified functionally relevant GO categories associated with the quantitative trait, chill coma recovery time, in the unrelated, sequenced inbred lines of the Drosophila melanogaster Genetic Reference Panel.

A complete tool set for molecular QTL discovery and analysis

Population scale studies combining genetic information with molecular phenotypes (for example, gene expression) have become a standard to dissect the effects of genetic variants onto organismal phenotypes. These kinds of data sets require powerful, fast and versatile methods able to discover molecular Quantitative Trait Loci (molQTL). Here we propose such a solution, QTLtools, a modular framework that contains multiple new and well-established methods to prepare the data, to discover proximal and distal molQTLs and, finally, to integrate them with GWAS variants and functional annotations of the genome. We demonstrate its utility by performing a complete expression QTL study in a few easy-to-perform steps. QTLtools is open source and available at https://qtltools.github.io/qtltools/.

Analysis of molecular quantitative trait loci (molQTL) can help interpret genome-wide association studies and requires efficient approaches to correct for multiple testing. This study describes a bioinformatics toolkit called QTLtool that can handle large data sets and quickly perform multiple types of molQTL analyses.

Antagonistic genetic correlations for milking traits within the genome of dairy cattle

Genome-wide association studies can be applied to identify useful SNPs associated with complex traits. Furthermore, regional genomic mapping can be used to estimate regional variance and clarify the genomic relationships within and outside regions but has not previously been applied to milk traits in cattle. We applied both single SNP analysis and regional genomic mapping to investigate SNPs or regions associated with milk yield traits in dairy cattle. The de-regressed breeding values of three traits, total yield (kg) of milk (MLK), fat (FAT), and protein (PRT) in 305 days, from 2,590 Holstein sires in Japan were analyzed. All sires were genotyped with 40,646 single-nucleotide polymorphism (SNP) markers. A genome-wide significant region (P < 0.01) common to all three traits was identified by regional genomic mapping on chromosome (BTA) 14. In contrast, single SNP analysis identified significant SNPs only for MLK and FAT (P < 0.01), but not PRT in the same region. Regional genomic mapping revealed an additional significant region (P < 0.01) for FAT on BTA5 that was not identified by single SNP analysis. The additive whole-genomic effects estimated in the regional genomic mapping analysis for the three traits were positively correlated with one another (0.830–0.924). However, the regional genomic effects obtained by using a window size of 20 SNPs for FAT on BTA14 were negatively correlated (P < 0.01) with the regional genomic effect for MLK (–0.940) and PRT (–0.878). The BTA14 regional effect for FAT also showed significant negative correlations (P < 0.01) with the whole genomic effects for MLK (–0.153), FAT (–0.172), and PRT (–0.181). These negative genomic correlations between loci are consistent with the negative linkage disequilibrium expected for traits under directional selection. Such antagonistic correlations may hamper the fixation of the FAT increasing alleles on BTA14. In summary, regional genomic mapping found more regions associated with milk production traits than did single SNP analysis. In addition, the existence of non-zero covariances between regional and whole genomic effects may influence the detection of regional effects, and antagonistic correlations could hamper the fixation of major genes under intensive selection.

Genomic regions underlying susceptibility to bovine tuberculosis in Holstein-Friesian cattle

BACKGROUND

The significant social and economic loss as a result of bovine tuberculosis (bTB) presents a continuous challenge to cattle industries in the UK and worldwide. However, host genetic variation in cattle susceptibility to bTB provides an opportunity to select for resistant animals and further understand the genetic mechanisms underlying disease dynamics.

METHODS

The present study identified genomic regions associated with susceptibility to bTB using genome-wide association (GWA), regional heritability mapping (RHM) and chromosome association approaches. Phenotypes comprised de-regressed estimated breeding values of 804 Holstein-Friesian sires and pertained to three bTB indicator traits: i) positive reactors to the skin test with positive post-mortem examination results (phenotype 1); ii) positive reactors to the skin test regardless of post-mortem examination results (phenotype 2) and iii) as in (ii) plus non-reactors and inconclusive reactors to the skin tests with positive post-mortem examination results (phenotype 3). Genotypes based on the 50 K SNP DNA array were available and a total of 34,874 SNPs remained per animal after quality control.

RESULTS

The estimated polygenic heritability for susceptibility to bTB was 0.26, 0.37 and 0.34 for phenotypes 1, 2 and 3, respectively. GWA analysis identified a putative SNP on Bos taurus autosomes (BTA) 2 associated with phenotype 1, and another on BTA 23 associated with phenotype 2. Genomic regions encompassing these SNPs were found to harbour potentially relevant annotated genes. RHM confirmed the effect of these genomic regions and identified new regions on BTA 18 for phenotype 1 and BTA 3 for phenotypes 2 and 3. Heritabilities of the genomic regions ranged between 0.05 and 0.08 across the three phenotypes. Chromosome association analysis indicated a major role of BTA 23 on susceptibility to bTB.

CONCLUSION

Genomic regions and candidate genes identified in the present study provide an opportunity to further understand pathways critical to cattle susceptibility to bTB and enhance genetic improvement programmes aiming at controlling and eradicating the disease.

Improved Genetic Profiling of Anthropometric Traits Using a Big Data Approach

Genome-wide association studies (GWAS) promised to translate their findings into clinically beneficial improvements of patient management by tailoring disease management to the individual through the prediction of disease risk. However, the ability to translate genetic findings from GWAS into predictive tools that are of clinical utility and which may inform clinical practice has, so far, been encouraging but limited. Here we propose to use a more powerful statistical approach, the use of which has traditionally been limited due to computational requirements and lack of sufficiently large individual level genotyped cohorts, but which improve the prediction of multiple medically relevant phenotypes using the same panel of SNPs. As a proof of principle, we used a shared panel of 319,038 common SNPs with MAF > 0.05 to train the prediction models in 114,264 unrelated White-British individuals for height and four obesity related traits (body mass index, basal metabolic rate, body fat percentage, and waist-to-hip ratio). We obtained prediction accuracies that ranged between 46% and 75% of the maximum achievable given the captured heritable component. For height, this represents an improvement in prediction accuracy of up to 68% (184% more phenotypic variance explained) over SNPs reported to be robustly associated with height in a previous GWAS meta-analysis of similar size. Across-population predictions in White non-British individuals were similar to those in White-British whilst those in Asian and Black individuals were informative but less accurate. We estimate that the genotyping of circa 500,000 unrelated individuals will yield predictions between 66% and 82% of the SNP-heritability captured by common variants in our array. Prediction accuracies did not improve when including rarer SNPs or when fitting multiple traits jointly in multivariate models.

Evaluating the contribution of genetics and familial shared environment to common disease using the UK Biobank

Genome-wide association studies have detected many loci underlying susceptibility to disease, but most of the genetic factors that contribute to disease susceptibility remain unknown. Here we provide evidence that part of the 'missing heritability' can be explained by an overestimation of heritability. We estimated the heritability of 12 complex human diseases using family history of disease in 1,555,906 individuals of white ancestry from the UK Biobank. Estimates using simple family-based statistical models were inflated on average by ∼47% when compared with those from structural equation modeling (SEM), which specifically accounted for shared familial environmental factors. In addition, heritabilities estimated using SNP data explained an average of 44.2% of the simple family-based estimates across diseases and an average of 57.3% of the SEM-estimated heritabilities, accounting for almost all of the SEM heritability for hypertension. Our results show that both genetics and familial environment make substantial contributions to familial clustering of disease.

Evidence for sex-specific genetic architectures across a spectrum of human complex traits

BACKGROUND

Sex differences are a common feature of human traits; however, the role sex determination plays in human genetic variation remains unclear. The presence of gene-by-sex (GxS) interactions implies that trait genetic architecture differs between men and women. Here, we show that GxS interactions and genetic heterogeneity among sexes are small but common features of a range of high-level complex traits.

RESULTS

We analyzed 19 complex traits measured in 54,040 unrelated men and 59,820 unrelated women from the UK Biobank cohort to estimate autosomal genetic correlations and heritability differences between men and women. For 13 of the 19 traits examined, there is evidence that the trait measured is genetically different between males and females. We find that estimates of genetic correlations, based on ~114,000 unrelated individuals and ~19,000 related individuals from the same cohort, are largely consistent. Genetic predictors using a sex-specific model that incorporated GxS interactions led to a relative improvement of up to 4 % (mean 1.4 % across all relevant phenotypes) over those provided by a sex-agnostic model. This further supports the hypothesis of the presence of sexual genetic heterogeneity across high-level phenotypes.

CONCLUSIONS

The sex-specific environment seems to play a role in changing genotype expression across a range of human complex traits. Further studies of GxS interactions for high-level human traits may shed light on the molecular mechanisms that lead to biological differences between men and women. However, this may be a challenging endeavour due to the likely small effects of the interactions at individual loci.

JAM: A Scalable Bayesian Framework for Joint Analysis of Marginal SNP Effects

Recently, large scale genome-wide association study (GWAS) meta-analyses have boosted the number of known signals for some traits into the tens and hundreds. Typically, however, variants are only analysed one-at-a-time. This complicates the ability of fine-mapping to identify a small set of SNPs for further functional follow-up. We describe a new and scalable algorithm, joint analysis of marginal summary statistics (JAM), for the re-analysis of published marginal summary statistics under joint multi-SNP models. The correlation is accounted for according to estimates from a reference dataset, and models and SNPs that best explain the complete joint pattern of marginal effects are highlighted via an integrated Bayesian penalized regression framework. We provide both enumerated and Reversible Jump MCMC implementations of JAM and present some comparisons of performance. In a series of realistic simulation studies, JAM demonstrated identical performance to various alternatives designed for single region settings. In multi-region settings, where the only multivariate alternative involves stepwise selection, JAM offered greater power and specificity. We also present an application to real published results from MAGIC (meta-analysis of glucose and insulin related traits consortium) - a GWAS meta-analysis of more than 15,000 people. We re-analysed several genomic regions that produced multiple significant signals with glucose levels 2 hr after oral stimulation. Through joint multivariate modelling, JAM was able to formally rule out many SNPs, and for one gene, ADCY5, suggests that an additional SNP, which transpired to be more biologically plausible, should be followed up with equal priority to the reported index.

Genetic determination of height-mediated mate choice

BACKGROUND

Numerous studies have reported positive correlations among couples for height. This suggests that humans find individuals of similar height attractive. However, the answer to whether the choice of a mate with a similar phenotype is genetically or environmentally determined has been elusive.

RESULTS

Here we provide an estimate of the genetic contribution to height choice in mates in 13,068 genotyped couples. Using a mixed linear model we show that 4.1% of the variation in the mate height choice is determined by a person's own genotype, as expected in a model where one's height determines the choice of mate height. Furthermore, the genotype of an individual predicts their partners' height in an independent dataset of 15,437 individuals with 13% accuracy, which is 64% of the theoretical maximum achievable with a heritability of 0.041. Theoretical predictions suggest that approximately 5% of the heritability of height is due to the positive covariance between allelic effects at different loci, which is caused by assortative mating. Hence, the coupling of alleles with similar effects could substantially contribute to the missing heritability of height.

CONCLUSIONS

These estimates provide new insight into the mechanisms that govern mate choice in humans and warrant the search for the genetic causes of choice of mate height. They have important methodological implications and contribute to the missing heritability debate.