Sparse Linear Mixed Model Research Articles

Genetic variation within a stick insect species associated with community-level traits.

Phenotypic variation within species can affect the ecological dynamics of populations and communities. Characterizing the genetic variation underlying such effects can help parse the roles of genetic evolution and plasticity in 'eco-evolutionary dynamics' and inform how genetic variation may shape patterns of evolution. Here we employ genome-wide association (GWA) methods in Timema cristinae stick insects and their co-occurring arthropod communities to identify genetic variation associated with community-level traits. Previous studies have shown that maladaptation (i.e., imperfect crypsis) of T. cristinae can reduce the abundance and species richness of other arthropods due to an increase in bird predation. Whether genetic variation that is independent from crypsis has similar effects is unknown and was tested here using genome-wide genotyping-by-sequencing data of stick insects, arthropod community information, and GWA mapping with Bayesian sparse linear mixed models. We find associations between genetic variation in stick insects and arthropod community traits. However, these associations disappeared when host-plant traits are accounted for. We thus use path analysis to disentangle interrelationships among stick-insect genetic variation, host-plant traits and community traits. This revealed that host-plant size has large effects on arthropod communities, while genetic variation in stick insects has a smaller, but still significant effect. Our findings demonstrate that: (1) genetic variation in a species can be associated with community-level traits, but that (2) interrelationships among multiple factors may need to be analyzed to disentangle whether such associations represent causal relationships. This work helps to build a framework for genomic studies of eco-evolutionary dynamics.

Methods for mediation analysis with high-dimensional DNA methylation data: Possible choices and comparisons.

Epigenetic researchers often evaluate DNA methylation as a potential mediator of the effect of social/environmental exposures on a health outcome. Modern statistical methods for jointly evaluating many mediators have not been widely adopted. We compare seven methods for high-dimensional mediation analysis with continuous outcomes through both diverse simulations and analysis of DNAm data from a large multi-ethnic cohort in the United States, while providing an R package for their seamless implementation and adoption. Among the considered choices, the best-performing methods for detecting active mediators in simulations are the Bayesian sparse linear mixed model (BSLMM) and high-dimensional mediation analysis (HDMA); while the preferred methods for estimating the global mediation effect are high-dimensional linear mediation analysis (HILMA) and principal component mediation analysis (PCMA). We provide guidelines for epigenetic researchers on choosing the best method in practice and offer suggestions for future methodological development.

Whole‐genome sequencing identifies interferon-induced protein IFI6/IFI27-like as a strong candidate gene for VNN resistance in European sea bass

BackgroundViral nervous necrosis (VNN) is a major disease that affects European sea bass, and understanding the biological mechanisms that underlie VNN resistance is important for the welfare of farmed fish and sustainability of production systems. The aim of this study was to identify genomic regions and genes that are associated with VNN resistance in sea bass.ResultsWe generated a dataset of 838,451 single nucleotide polymorphisms (SNPs) identified from whole-genome sequencing (WGS) in the parental generation of two commercial populations (A: 2371 individuals and B: 3428 individuals) of European sea bass with phenotypic records for binary survival in a VNN challenge. For each population, three cohorts were submitted to a red-spotted grouper nervous necrosis virus (RGNNV) challenge by immersion and genotyped on a 57K SNP chip. After imputation of WGS SNPs from their parents, quantitative trait loci (QTL) were mapped using a Bayesian sparse linear mixed model (BSLMM). We found several QTL regions that were specific to one of the populations on different linkage groups (LG), and one 127-kb QTL region on LG12 that was shared by both populations and included the genes ZDHHC14, which encodes a palmitoyltransferase, and IFI6/IFI27-like, which encodes an interferon-alpha induced protein. The most significant SNP in this QTL region was only 1.9 kb downstream of the coding sequence of the IFI6/IFI27-like gene. An unrelated population of four large families was used to validate the effect of the QTL. Survival rates of susceptible genotypes were 40.6% and 45.4% in populations A and B, respectively, while that of the resistant genotype was 66.2% in population B and 78% in population A.ConclusionsWe have identified a genomic region that carries a major QTL for resistance to VNN and includes the ZDHHC14 and IFI6/IFI27-like genes. The potential involvement of the interferon pathway, a well-known anti-viral defense mechanism in several organisms (chicken, human, or fish), in survival to VNN infection is of particular interest. Our results can lead to major improvements for sea bass breeding programs through marker-assisted genomic selection to obtain more resistant fish.

A novel sparse linear mixed model for multi-source mixed-frequency data fusion in telemedicine

Digital health and telemonitoring have resulted in a wealth of information to be collected to monitor, manage, and improve human health. The multi-source mixed-frequency health data overwhelm the modeling capacity of existing statistical and machine learning models, due to many challenging properties. Although predictive analytics for big health data plays an important role in telemonitoring, there is a lack of rigorous prediction model that can automatically predicts patients’ health conditions, e.g., Disease Severity Indicators (DSIs), from multi-source mixed-frequency data. Sleep disorder is a prevalent cardiac syndrome that is characterized by abnormal respiratory patterns during sleep. Although wearable devices are available to administrate sleep studies at home, the manual scoring process to generate the DSI remains a bottleneck in automated monitoring and diagnosis of sleep disorder. To address the multi-fold challenges for precise prediction of the DSI from high-dimensional multi-source mixed-frequency data in sleep disorder, we propose a sparse linear mixed model that combines the modified Cholesky decomposition with group lasso penalties to enable joint group selection of fixed effects and random effects. A novel Expectation Maximization (EM) algorithm integrated with an efficient Majorization Maximization (MM) algorithm is developed for model estimation of the proposed sparse linear mixed model with group variable selection. The proposed method was applied to the SHHS data for telemonitoring and diagnosis of sleep disorder and found that a few significant feature groups that are consistent with prior medical studies on sleep disorder. The proposed method also outperformed a few benchmark methods with the highest prediction accuracy.

Genomic prediction in pigs using data from a commercial crossbred population: insights from the Duroc x (Landrace x Yorkshire) three-way crossbreeding system

BackgroundGenomic selection is widely applied for genetic improvement in livestock crossbreeding systems to select excellent nucleus purebred (PB) animals and to improve the performance of commercial crossbred (CB) animals. Most current predictions are based solely on PB performance. Our objective was to explore the potential application of genomic selection of PB animals using genotypes of CB animals with extreme phenotypes in a three-way crossbreeding system as the reference population. Using real genotyped PB as ancestors, we simulated the production of 100,000 pigs for a Duroc x (Landrace x Yorkshire) DLY crossbreeding system. The predictive performance of breeding values of PB animals for CB performance using genotypes and phenotypes of (1) PB animals, (2) DLY animals with extreme phenotypes, and (3) random DLY animals for traits of different heritabilities ({h}^{2} = 0.1, 0.3, and 0.5) was compared across different reference population sizes (500 to 6500) and prediction models (genomic best linear unbiased prediction (GBLUP) and Bayesian sparse linear mixed model (BSLMM)).ResultsUsing a reference population consisting of CB animals with extreme phenotypes showed a definite predictive advantage for medium- and low-heritability traits and, in combination with the BSLMM model, significantly improved selection response for CB performance. For high-heritability traits, the predictive performance of a reference population of extreme CB phenotypes was comparable to that of PB phenotypes when the effect of the genetic correlation between PB and CB performance ({r}_{pc}) on the accuracy obtained with a PB reference population was considered, and the former could exceed the latter if the reference size was large enough. For the selection of the first and terminal sires in a three-way crossbreeding system, prediction using extreme CB phenotypes outperformed the use of PB phenotypes, while the optimal design of the reference group for the first dam depended on the percentage of individuals from the corresponding breed that the PB reference data comprised and on the heritability of the target trait.ConclusionsA commercial crossbred population is promising for the design of the reference population for genomic prediction, and selective genotyping of CB animals with extreme phenotypes has the potential for maximizing genetic improvement for CB performance in the pig industry.

Genetic basis of growth reaction to drought stress differs in contrasting high-latitude treeline ecotones of a widespread conifer.

Climate change is increasing the frequency and intensity of drought events in many boreal forests. Trees are sessile organisms with a long generation time, which makes them vulnerable to fast climate change and hinders fast adaptations. Therefore, it is important to know how forests cope with drought stress and to explore the genetic basis of these reactions. We investigated three natural populations of white spruce (Picea glauca) in Alaska, located at one drought-limited and two cold-limited treelines with a paired plot design of one forest and one treeline plot. We obtained individual increment cores from 458 trees and climate data to assess dendrophenotypes, in particular the growth reaction to drought stress. To explore the genetic basis of these dendrophenotypes, we genotyped the individual trees at 3000 single nucleotide polymorphisms in candidate genes and performed genotype-phenotype association analysis using linear mixed models and Bayesian sparse linear mixed models. Growth reaction to drought stress differed in contrasting treeline populations. Therefore, the populations are likely to be unevenly affected by climate change. We identified 40 genes associated with dendrophenotypic traits that differed among the treeline populations. Most genes were identified in the drought-limited site, indicating comparatively strong selection pressure of drought-tolerant phenotypes. Contrasting patterns of drought-associated genes among sampled sites and in comparison to Canadian populations in a previous study suggest that drought adaptation acts on a local scale. Our results highlight genes that are associated with wood traits which in turn are critical for the establishment and persistence of future forests under climate change.

A polygenic architecture with habitat-dependent effects underlies ecological differentiation in Silene.

Summary Ecological differentiation can drive speciation but it is unclear how the genetic architecture of habitat‐dependent fitness contributes to lineage divergence. We investigated the genetic architecture of cumulative flowering, a fitness component, in second‐generation hybrids between Silene dioica and Silene latifolia transplanted into the natural habitat of each species.We used reduced‐representation sequencing and Bayesian sparse linear mixed models (BSLMMs) to analyze the genetic control of cumulative flowering in each habitat.Our results point to a polygenic architecture of cumulative flowering. Allelic effects were mostly beneficial or deleterious in one habitat and neutral in the other. Positive‐effect alleles often were derived from the native species, whereas negative‐effect alleles, at other loci, tended to originate from the non‐native species.We conclude that ecological differentiation is governed and maintained by many loci with small, habitat‐dependent effects consistent with conditional neutrality. This pattern may result from differences in selection targets in the two habitats and from environmentally dependent deleterious load. Our results further suggest that selection for native alleles and against non‐native alleles acts as a barrier to gene flow between species.

Genome-Wide Association Analysis and Genomic Prediction of Thyroglobulin Plasma Levels.

Thyroglobulin (Tg) is an iodoglycoprotein produced by thyroid follicular cells which acts as an essential substrate for thyroid hormone synthesis. To date, only one genome-wide association study (GWAS) of plasma Tg levels has been performed by our research group. Utilizing recent advancements in computation and modeling, we apply a Bayesian approach to the probabilistic inference of the genetic architecture of Tg. We fitted a Bayesian sparse linear mixed model (BSLMM) and a frequentist linear mixed model (LMM) of 7,289,083 variants in 1096 healthy European-ancestry participants of the Croatian Biobank. Meta-analysis with two independent cohorts (total n = 2109) identified 83 genome-wide significant single nucleotide polymorphisms (SNPs) within the ST6GAL1 gene (). BSLMM revealed additional association signals on chromosomes 1, 8, 10, and 14. For ST6GAL1 and the newly uncovered genes, we provide physiological and pathophysiological explanations of how their expression could be associated with variations in plasma Tg levels. We found that the SNP-heritability of Tg is 17% and that 52% of this variation is due to a small number of 16 variants that have a major effect on Tg levels. Our results suggest that the genetic architecture of plasma Tg is not polygenic, but influenced by a few genes with major effects.

Sparse latent factor regression models for genome-wide and epigenome-wide association studies.

Association of phenotypes or exposures with genomic and epigenomic data faces important statistical challenges. One of these challenges is to account for variation due to unobserved confounding factors, such as individual ancestry or cell-type composition in tissues. This issue can be addressed with penalized latent factor regression models, where penalties are introduced to cope with high dimension in the data. If a relatively small proportion of genomic or epigenomic markers correlate with the variable of interest, sparsity penalties may help to capture the relevant associations, but the improvement over non-sparse approaches has not been fully evaluated yet. Here, we present least-squares algorithms that jointly estimate effect sizes and confounding factors in sparse latent factor regression models. In simulated data, sparse latent factor regression models generally achieved higher statistical performance than other sparse methods, including the least absolute shrinkage and selection operator and a Bayesian sparse linear mixed model. In generative model simulations, statistical performance was slightly lower (while being comparable) to non-sparse methods, but in simulations based on empirical data, sparse latent factor regression models were more robust to departure from the model than the non-sparse approaches. We applied sparse latent factor regression models to a genome-wide association study of a flowering trait for the plant Arabidopsis thaliana and to an epigenome-wide association study of smoking status in pregnant women. For both applications, sparse latent factor regression models facilitated the estimation of non-null effect sizes while overcoming multiple testing issues. The results were not only consistent with previous discoveries, but they also pinpointed new genes with functional annotations relevant to each application.

Sparse linear mixed model selection via streamlined variational Bayes

Linear mixed models are a versatile statistical tool to study data by accounting for fixed effects and random effects from multiple sources of variability. In many situations, a large number of candidate fixed effects is available and it is of interest to select a parsimonious subset of those being effectively relevant for predicting the response variable. Variational approximations facilitate fast approximate Bayesian inference for the parameters of a variety of statistical models, including linear mixed models. However, for models having a high number of fixed or random effects, simple application of standard variational inference principles does not lead to fast approximate inference algorithms, due to the size of model design matrices and inefficient treatment of sparse matrix problems arising from the required approximating density parameters updates. We illustrate how recently developed streamlined variational inference procedures can be generalized to make fast and accurate inference for the parameters of linear mixed models with nested random effects and global-local priors for Bayesian fixed effects selection. Our variational inference algorithms achieve convergence to the same optima of their standard implementations, although with significantly lower computational effort, memory usage and time, especially for large numbers of random effects. Using simulated and real data examples, we assess the quality of automated procedures for fixed effects selection that are free from hyperparameters tuning and only rely upon variational posterior approximations. Moreover, we show high accuracy of variational approximations against model fitting via Markov Chain Monte Carlo sampling.

Performances of Adaptive MultiBLUP, Bayesian regressions, and weighted-GBLUP approaches for genomic predictions in Belgian Blue beef cattle

BackgroundGenomic selection has been successfully implemented in many livestock and crop species. The genomic best linear unbiased predictor (GBLUP) approach, assigning equal variance to all SNP effects, is one of the reference methods. When large-effect variants contribute to complex traits, it has been shown that genomic prediction methods that assign a higher variance to subsets of SNP effects can achieve higher prediction accuracy. We herein compared the efficiency of several such approaches, including the Adaptive MultiBLUP (AM-BLUP) that uses local genomic relationship matrices (GRM) to automatically identify and weight genomic regions with large effects, to predict genetic merit in Belgian Blue beef cattle.ResultsWe used a population of approximately 10,000 genotyped cows and their phenotypes for 14 traits, mostly related to muscular development and body dimensions. According to the trait, we found that 4 to 25% of the genetic variance could be associated with 2 to 12 genomic regions harbouring large-effect variants. Noteworthy, three previously identified recessive deleterious variants presented heterozygote advantage and were among the most significant SNPs for several traits. The AM-BLUP resulted in increased reliability of genomic predictions compared to GBLUP (+ 2%), but Bayesian methods proved more efficient (+ 3%). Overall, the reliability gains remained thus limited although higher gains were observed for skin thickness, a trait affected by two genomic regions having particularly large effects. Higher accuracies than those from the original AM-BLUP were achieved when applying the Bayesian Sparse Linear Mixed Model to pre-select groups of SNPs with large effects and subsequently use their estimated variance to build a weighted GRM. Finally, the single-step GBLUP performed best and could be further improved (+ 3% prediction accuracy) by using these weighted GRM.ConclusionsThe AM-BLUP is an attractive method to automatically identify and weight genomic regions with large effects on complex traits. However, the method was less accurate than Bayesian methods. Overall, weighted methods achieved modest accuracy gains compared to GBLUP. Nevertheless, the computational efficiency of the AM-BLUP might be valuable at higher marker density, including with whole-genome sequencing data. Furthermore, weighted GRM are particularly useful to account for large variance loci in the single-step GBLUP.

Accurate and Scalable Construction of Polygenic Scores in Large Biobank Data Sets

Accurate construction of polygenic scores (PGS) can enable early diagnosis of diseases and facilitate the development of personalized medicine. Accurate PGS construction requires prediction models that are both adaptive to different genetic architectures and scalable to biobank scale datasets with millions of individuals and tens of millions of genetic variants. Here, we develop such a method called Deterministic Bayesian Sparse Linear Mixed Model (DBSLMM). DBSLMM relies on a flexible modeling assumption on the effect size distribution to achieve robust and accurate prediction performance across a range of genetic architectures. DBSLMM also relies on a simple deterministic search algorithm to yield an approximate analytic estimation solution using summary statistics only. The deterministic search algorithm, when paired with further algebraic innovations, results in substantial computational savings. With simulations, we show that DBSLMM achieves scalable and accurate prediction performance across a range of realistic genetic architectures. We then apply DBSLMM to analyze 25 traits in UK Biobank. For these traits, compared to existing approaches, DBSLMM achieves an average of 2.03%-101.09% accuracy gain in internal cross-validations. In external validations on two separate datasets, including one from BioBank Japan, DBSLMM achieves an average of 14.74%-522.74% accuracy gain. In these real data applications, DBSLMM is 1.03-28.11 times faster and uses only 7.4%-24.8% of physical memory as compared to other multiple regression-based PGS methods. Overall, DBSLMM represents an accurate and scalable method for constructing PGS in biobank scale datasets.

Discovering weaker genetic associations guided by known associations

BackgroundThe current understanding of the genetic basis of complex human diseases is that they are caused and affected by many common and rare genetic variants. A considerable number of the disease-associated variants have been identified by Genome Wide Association Studies, however, they can explain only a small proportion of heritability. One of the possible reasons for the missing heritability is that many undiscovered disease-causing variants are weakly associated with the disease. This can pose serious challenges to many statistical methods, which seems to be only capable of identifying disease-associated variants with relatively stronger coefficients.ResultsIn order to help identify weaker variants, we propose a novel statistical method, Constrained Sparse multi-locus Linear Mixed Model (CS-LMM) that aims to uncover genetic variants of weaker associations by incorporating known associations as a prior knowledge in the model. Moreover, CS-LMM accounts for polygenic effects as well as corrects for complex relatednesses. Our simulation experiments show that CS-LMM outperforms other competing existing methods in various settings when the combinations of MAFs and coefficients reflect different scenarios in complex human diseases.ConclusionsWe also apply our method to the GWAS data of alcoholism and Alzheimer’s disease and exploratively discover several SNPs. Many of these discoveries are supported through literature survey. Furthermore, our association results strengthen the belief in genetic links between alcoholism and Alzheimer’s disease.

Examining the Impact of Imputation Errors on Fine-Mapping Using DNA Methylation QTL as a Model Trait.

Genetic variants disrupting DNA methylation at CpG dinucleotides (CpG-SNP) provide a set of known causal variants to serve as models to test fine-mapping methodology. We use 1716 CpG-SNPs to test three fine-mapping approaches (Bayesian imputation-based association mapping, Bayesian sparse linear mixed model, and the J-test), assessing the impact of imputation errors and the choice of reference panel by using both whole-genome sequence (WGS), and genotype array data on the same individuals (n = 1166). The choice of imputation reference panel had a strong effect on imputation accuracy, with the 1000 Genomes Project Phase 3 (1000G) reference panel (n = 2504 from 26 populations) giving a mean nonreference discordance rate between imputed and sequenced genotypes of 3.2% compared to 1.6% when using the Haplotype Reference Consortium (HRC) reference panel (n = 32,470 Europeans). These imputation errors had an impact on whether the CpG-SNP was included in the 95% credible set, with a difference of ∼23% and ∼7% between the WGS and the 1000G and HRC imputed datasets, respectively. All of the fine-mapping methods failed to reach the expected 95% coverage of the CpG-SNP. This is attributed to secondary cis genetic effects that are unable to be statistically separated from the CpG-SNP, and through a masking mechanism where the effect of the methylation disrupting allele at the CpG-SNP is hidden by the effect of a nearby SNP that has strong linkage disequilibrium with the CpG-SNP. The reduced accuracy in fine-mapping a known causal variant in a low-level biological trait with imputed genetic data has implications for the study of higher-order complex traits and disease.

Evolutionary genomics of gypsy moth populations sampled along a latitudinal gradient.

The European gypsy moth (Lymantria dispar L.) was first introduced to Massachusetts in 1869 and within 150years has spread throughout eastern North America. This large-scale invasion across a heterogeneous landscape allows examination of the genetic signatures of adaptation potentially associated with rapid geographical spread. We tested the hypothesis that spatially divergent natural selection has driven observed changes in three developmental traits that were measured in a common garden for 165 adult moths sampled from six populations across a latitudinal gradient covering the entirety of the range. We generated genotype data for 91,468 single nucleotide polymorphisms based on double digest restriction-site associated DNA sequencing and used these data to discover genome-wide associations for each trait, as well as to test for signatures of selection on the discovered architectures. Genetic structure across the introduced range of gypsy moth was low in magnitude (FST =0.069), with signatures of bottlenecks and spatial expansion apparent in the rare portion of the allele frequency spectrum. Results from applications of Bayesian sparse linear mixed models were consistent with the presumed polygenic architectures of each trait. Further analyses indicated spatially divergent natural selection acting on larval development time and pupal mass, with the linkage disequilibrium component of this test acting as the main driver of observed patterns. The populations most important for these signals were two range-edge populations established less than 30 generations ago. We discuss the importance of rapid polygenic adaptation to the ability of non-native species to invade novel environments.

Combining Sparse Group Lasso and Linear Mixed Model Improves Power to Detect Genetic Variants Underlying Quantitative Traits.

Genome-Wide association studies (GWAS), based on testing one single nucleotide polymorphism (SNP) at a time, have revolutionized our understanding of the genetics of complex traits. In GWAS, there is a need to consider confounding effects such as due to population structure, and take groups of SNPs into account simultaneously due to the “polygenic” attribute of complex quantitative traits. In this paper, we propose a new approach SGL-LMM that puts together sparse group lasso (SGL) and linear mixed model (LMM) for multivariate associations of quantitative traits. LMM, as has been often used in GWAS, controls for confounders, while SGL maintains sparsity of the underlying multivariate regression model. SGL-LMM first sets a fixed zero effect to learn the parameters of random effects using LMM, and then estimates fixed effects using SGL regularization. We present efficient algorithms for hyperparameter tuning and feature selection using stability selection. While controlling for confounders and constraining for sparse solutions, SGL-LMM also provides a natural framework for incorporating prior biological information into the group structure underlying the model. Results based on both simulated and real data show SGL-LMM outperforms previous approaches in terms of power to detect associations and accuracy of quantitative trait prediction.

Multiplex confounding factor correction for genomic association mapping with squared sparse linear mixed model.

Genome-wide Association Study has presented a promising way to understand the association between human genomes and complex traits. Many simple polymorphic loci have been shown to explain a significant fraction of phenotypic variability. However, challenges remain in the non-triviality of explaining complex traits associated with multifactorial genetic loci, especially considering the confounding factors caused by population structure, family structure, and cryptic relatedness. In this paper, we propose a Squared-LMM (LMM2) model, aiming to jointly correct population and genetic confounding factors. We offer two strategies of utilizing LMM2 for association mapping: 1) It serves as an extension of univariate LMM, which could effectively correct population structure, but consider each SNP in isolation. 2) It is integrated with the multivariate regression model to discover association relationship between complex traits and multifactorial genetic loci. We refer to this second model as sparse Squared-LMM (sLMM2). Further, we extend LMM2/sLMM2 by raising the power of our squared model to the LMMn/sLMMn model. We demonstrate the practical use of our model with synthetic phenotypic variants generated from genetic loci of Arabidopsis Thaliana. The experiment shows that our method achieves a more accurate and significant prediction on the association relationship between traits and loci. We also evaluate our models on collected phenotypes and genotypes with the number of candidate genes that the models could discover. The results suggest the potential and promising usage of our method in genome-wide association studies.

Variable selection in heterogeneous datasets: A truncated-rank sparse linear mixed model with applications to genome-wide association studies.

A fundamental and important challenge in modern datasets of ever increasing dimensionality is variable selection, which has taken on renewed interest recently due to the growth of biological and medical datasets with complex, non-i.i.d. structures. Naïvely applying classical variable selection methods such as the Lasso to such datasets may lead to a large number of false discoveries. Motivated by genome-wide association studies in genetics, we study the problem of variable selection for datasets arising from multiple subpopulations, when this underlying population structure is unknown to the researcher. We propose a unified framework for sparse variable selection that adaptively corrects for population structure via a low-rank linear mixed model. Most importantly, the proposed method does not require prior knowledge of sample structure in the data and adaptively selects a covariance structure of the correct complexity. Through extensive experiments, we illustrate the effectiveness of this framework over existing methods. Further, we test our method on three different genomic datasets from plants, mice, and human, and discuss the knowledge we discover with our method.

Improving Heritability Estimation by a Variable Selection Approach in Sparse High Dimensional Linear Mixed Models

SummaryMotivated by applications in neuroanatomy, we propose a novel methodology to estimate heritability, which corresponds to the proportion of phenotypic variance that can be explained by genetic factors. Since the phenotypic variations may be due to only a small fraction of the available genetic information, we propose an estimator of heritability that can be used in sparse linear mixed models. Since the real genetic architecture is in general unknown in practice, our method enables the user to determine whether the genetic effects are very sparse: in that case, we propose a variable selection approach to recover the support of these genetic effects before estimating heritability. Otherwise, we use a classical maximum likelihood approach. We apply our method, implemented in the R package EstHer that is available on the Comprehensive R Archive Network, on neuroanatomical data from the project IMAGEN.

A missense variant in the coil1A domain of the keratin 25 gene is associated with the dominant curly hair coat trait (Crd) in horse

BackgroundCurly horses present a variety of curl phenotypes that are associated with various degrees of curliness of coat, mane, tail and ear hairs. Their origin is still a matter of debate and several genetic hypotheses have been formulated to explain the diversity in phenotype, including the combination of autosomal dominant and recessive alleles. Our purpose was to map the autosomal dominant curly hair locus and identify the causal variant using genome-wide association study (GWAS) and whole-genome sequencing approaches.ResultsA GWAS was performed using a Bayesian sparse linear mixed model, based on 51 curly and 19 straight-haired French and North American horses from 13 paternal families genotyped on the Illumina EquineSNP50 BeadChip. A single strong signal was observed on equine chromosome 11, in a region that encompasses the type I keratin gene cluster. This region was refined by haplotype analysis to a segment including 36 genes, among which are 10 keratin genes (KRT-10, -12, -20, -23, -24, -25, -26, -27, -28, -222). To comprehensively identify candidate causal variants within all these genes, whole-genome sequences were obtained for one heterozygous curly stallion and its straight-haired son. Among the four non-synonymous candidate variants identified and validated in the curly region, only variant g.21891160G>A in the KRT25 gene (KRT25:p.R89H) was in perfect agreement with haplotype status in the whole pedigree. Genetic association was then confirmed by genotyping a larger population consisting of 353 horses. However, five discordant curly horses were observed, which carried neither the variant nor the main haplotype associated with curliness. Sequencing of KRT25 for two discordant horses did not identify any other deleterious variant, which suggests locus rather than allelic heterogeneity for the curly phenotype.ConclusionsWe identified the KRT25:p.R89H variant as responsible for the dominant curly trait, but a second dominant locus may also be involved in the shape of hairs within North American Curly horses.