Congruence of additive and non-additive effects on gene expression estimated from pedigree and SNP data

Dynamic patterns of gene expressional and regulatory variations in cotton heterosis

Purpose

Although the application of heterosis has significantly increased crop yield over the past century, the mechanisms underlying this phenomenon still remain obscure. Here, we applied transcriptome sequencing to unravel the impacts of parental expression differences and transcriptomic reprogramming in cotton heterosis.

Methods

A high-quality transcriptomic atlas covering 15 developmental stages and tissues was constructed for XZM2, an elite hybrid of upland cotton (Gossypium hirsutum L.), and its parental lines, CRI12 and J8891. This atlas allowed us to identify gene expression differences between the parents and to characterize the transcriptomic reprogramming that occurs in the hybrid.

Results

Our analysis revealed abundant gene expression differences between the parents, with pronounced tissue specificity; a total of 1,112 genes exhibited single-parent expression in at least one tissue. It also illuminated transcriptomic reprogramming in the hybrid XZM2, which included both additive and non-additive expression patterns. Coexpression networks between parents and hybrid constructed via weighted gene coexpression network analysis identified modules closely associated with fiber development. In particular, key regulatory hub genes involved in fiber development showed high-parent dominant or over dominant patterns in the hybrid, potentially driving the emergence of heterosis. Finally, high-depth resequencing data was generated and allele-specific expression patterns examined in the hybrid, enabling the dissection of cis- and trans-regulation contributions to the observed expression differences.

Conclusion

Parental transcriptional differences and transcriptomic reprogramming in the hybrid, especially the non-additive upregulation of key genes, play an important role in shaping heterosis. Collectively, these findings provide new insights into the molecular basis of heterosis in cotton.

Non-additive genetic components contribute significantly to population-wide gene expression variation

Gene expression variation, an essential step between genotype and phenotype, is collectively controlled by local (cis) and distant (trans) regulatory changes. Nevertheless, how these regulatory elements differentially influence gene expression variation remains unclear. Here, we bridge this gap by analyzing the transcriptomes of a large diallel panel consisting of 323 unique hybrids originating from genetically divergent Saccharomyces cerevisiae isolates. Our analysis across 5,087 transcript abundance traits showed that non-additive components account for 36% of the gene expression variance on average. By comparing allele-specific read counts in parent-hybrid trios, we found that trans-regulatory changes underlie the majority of gene expression variation in the population. Remarkably, most cis-regulatory variations are also exaggerated or attenuated by additional trans effects. Overall, we showed that the transcriptome is globally buffered at the genetic level mainly due to trans-regulatory variation in the population.

Dominance is common in mammals and is associated with trans-acting gene expression and alternative splicing

BACKGROUND

Dominance and other non-additive genetic effects arise from the interaction between alleles, and historically these phenomena play a major role in quantitative genetics. However, most genome-wide association studies (GWAS) assume alleles act additively.

RESULTS

We systematically investigate both dominance-here representing any non-additive within-locus interaction-and additivity across 574 physiological and gene expression traits in three mammalian stocks: F2 intercross pigs, rat heterogeneous stock, and mice heterogeneous stock. Dominance accounts for about one quarter of heritable variance across all physiological traits in all species. Hematological and immunological traits exhibit the highest dominance variance, possibly reflecting balancing selection in response to pathogens. Although most quantitative trait loci (QTLs) are detectable as additive QTLs, we identify 154, 64, and 62 novel dominance QTLs in pigs, rats, and mice respectively that are undetectable as additive QTLs. Similarly, even though most cis-acting expression QTLs are additive, gene expression exhibits a large fraction of dominance variance, and trans-acting eQTLs are enriched for dominance. Genes causal for dominance physiological QTLs are less likely to be physically linked to their QTLs but instead act via trans-acting dominance eQTLs. In addition, thousands of eQTLs are associated with alternatively spliced isoforms with complex additive and dominant architectures in heterogeneous stock rats, suggesting a possible mechanism for dominance.

CONCLUSIONS

Although heritability is predominantly additive, many mammalian genetic effects are dominant and likely arise through distinct mechanisms. It is therefore advantageous to consider both additive and dominance effects in GWAS to improve power and uncover causality.

Non-additive genetic components contribute significantly to population-wide gene expression variation

Gene expression variation, an essential step between genomic variation and phenotypic landscape, is collectively controlled by local (cis) and distant (trans) regulatory changes. Nevertheless, how these regulatory elements differentially influence the heritability of expression traits remains unclear. Here, we bridge this gap by analyzing the transcriptomes of a large diallel panel consisting of 323 unique hybrids originated from genetically divergent yeast isolates. We estimated the broad- and narrow-sense heritability across 5,087 transcript abundance traits and showed that non-additive components account for 36% of the phenotypic variance on average. By comparing allelic expression ratios in the hybrid and the corresponding parental pair, we identified regulatory changes in 25% of all cases, with a majority acting in trans. We further showed that trans-regulation could underlie coordinated expression variation across highly connected genes, resulting in significantly higher non-additive variance and most likely in some of the missing heritability of gene expression traits.

Multiple causal variants underlie genetic associations in humans

Associations between genetic variation and traits are often in noncoding regions with strong linkage disequilibrium (LD), where a single causal variant is assumed to underlie the association. We applied a massively parallel reporter assay (MPRA) to functionally evaluate genetic variants in high, local LD for independent cis-expression quantitative trait loci (eQTL). We found that 17.7% of eQTLs exhibit more than one major allelic effect in tight LD. The detected regulatory variants were highly and specifically enriched for activating chromatin structures and allelic transcription factor binding. Integration of MPRA profiles with eQTL/complex trait colocalizations across 114 human traits and diseases identified causal variant sets demonstrating how genetic association signals can manifest through multiple, tightly linked causal variants.

Systemic lupus erythematosus as a genetic disease

Systemic lupus erythematosus is the prototypical systemic autoimmune disease, as it is characterized both by protean multi-organ system manifestations and by the uniform presence of pathogenic autoantibodies directed against components of the nucleus. Prior to the modern genetic era, the diverse clinical manifestations of SLE suggested to many that SLE patients were unlikely to share a common genetic risk basis. However, modern genetic studies have revealed that SLE usually arises when an environmental exposure occurs in an individual with a collection of genetic risk variants passing a liability threshold. Here, we summarize the current state of the field aimed at: (1) understanding the genetic architecture of this complex disease, (2) synthesizing how this genetic risk architecture impacts cellular and molecular disease pathophysiology, (3) providing illustrative examples that highlight the rich complexity of the pathobiology of this prototypical autoimmune disease and (4) communicating this complex etiopathogenesis to patients.

The Association Between Genomic Heterozygosity and Carcass Merit in Cattle

The objective of the present study was to quantify the association between both pedigree and genome-based measures of global heterozygosity and carcass traits, and to identify single nucleotide polymorphisms (SNPs) exhibiting non-additive associations with these traits. The carcass traits of interest were carcass weight (CW), carcass conformation (CC) and carcass fat (CF). To define the genome-based measures of heterozygosity, and to quantify the non-additive associations between SNPs and the carcass traits, imputed, high-density genotype data, comprising of 619,158 SNPs, from 27,213 cattle were used. The correlations between the pedigree-based heterosis coefficient and the three defined genomic measures of heterozygosity ranged from 0.18 to 0.76. The associations between the different measures of heterozygosity and the carcass traits were biologically small, with positive associations for CW and CC, and negative associations for CF. Furthermore, even after accounting for the pedigree-based heterosis coefficient of an animal, part of the remaining variability in some of the carcass traits could be captured by a genomic heterozygosity measure. This signifies that the inclusion of both a heterosis coefficient based on pedigree information and a genome-based measure of heterozygosity could be beneficial to limiting bias in predicting additive genetic merit. Finally, one SNP located on Bos taurus (BTA) chromosome number 5 demonstrated a non-additive association with CW. Furthermore, 182 SNPs (180 SNPs on BTA 2 and two SNPs on BTA 21) demonstrated a non-additive association with CC, while 231 SNPs located on BTA 2, 5, 11, 13, 14, 18, 19 and 21 demonstrated a non-additive association with CF. Results demonstrate that heterozygosity both at a global level and at the level of individual loci contribute little to the variability in carcass merit.

Combinatorial patterns of gene expression changes contribute to variable expressivity of the developmental delay-associated 16p12.1 deletion

BACKGROUND

Recent studies have suggested that individual variants do not sufficiently explain the variable expressivity of phenotypes observed in complex disorders. For example, the 16p12.1 deletion is associated with developmental delay and neuropsychiatric features in affected individuals, but is inherited in > 90% of cases from a mildly-affected parent. While children with the deletion are more likely to carry additional "second-hit" variants than their parents, the mechanisms for how these variants contribute to phenotypic variability are unknown.

METHODS

We performed detailed clinical assessments, whole-genome sequencing, and RNA sequencing of lymphoblastoid cell lines for 32 individuals in five large families with multiple members carrying the 16p12.1 deletion. We identified contributions of the 16p12.1 deletion and "second-hit" variants towards a range of expression changes in deletion carriers and their family members, including differential expression, outlier expression, alternative splicing, allele-specific expression, and expression quantitative trait loci analyses.

RESULTS

We found that the deletion dysregulates multiple autism and brain development genes such as FOXP1, ANK3, and MEF2. Carrier children also showed an average of 5323 gene expression changes compared with one or both parents, which matched with 33/39 observed developmental phenotypes. We identified significant enrichments for 13/25 classes of "second-hit" variants in genes with expression changes, where 4/25 variant classes were only enriched when inherited from the noncarrier parent, including loss-of-function SNVs and large duplications. In 11 instances, including for ZEB2 and SYNJ1, gene expression was synergistically altered by both the deletion and inherited "second-hits" in carrier children. Finally, brain-specific interaction network analysis showed strong connectivity between genes carrying "second-hits" and genes with transcriptome alterations in deletion carriers.

CONCLUSIONS

Our results suggest a potential mechanism for how "second-hit" variants modulate expressivity of complex disorders such as the 16p12.1 deletion through transcriptomic perturbation of gene networks important for early development. Our work further shows that family-based assessments of transcriptome data are highly relevant towards understanding the genetic mechanisms associated with complex disorders.

Multi-tissue transcriptome-wide association studies

A transcriptome-wide association study (TWAS) attempts to identify disease associated genes by imputing gene expression into a genome-wide association study (GWAS) using an expression quantitative trait loci (eQTL) data set and then testing for associations with a trait of interest. Regulatory processes may be shared across related tissues and one natural extension of TWAS is harnessing cross-tissue correlation in gene expression to improve prediction accuracy. Here, we studied multi-tissue extensions of lasso regression and random forests (RF), joint lasso and RF-MTL (multi-task learning RF), respectively. We found that, on our chosen eQTL data set, multi-tissue methods were generally more accurate than their single-tissue counterparts, with RF-MTL performing the best. Simulations showed that these benefits generally translated into more associated genes identified, although highlighted that joint lasso had a tendency to erroneously identify genes in one tissue if there existed an eQTL signal for that gene in another. Applying the four methods to a type 1 diabetes GWAS, we found that multi-tissue methods found more unique associated genes for most of the tissues considered. We conclude that multi-tissue methods are competitive and, for some cell types, superior to single-tissue approaches and hold much promise for TWAS studies.

Autozygosity mapping and time-to-spontaneous delivery in Norwegian parent-offspring trios

Parental genetic relatedness may lead to adverse health and fitness outcomes in the offspring. However, the degree to which it affects human delivery timing is unknown. We use genotype data from ≃25 000 parent-offspring trios from the Norwegian Mother, Father and Child Cohort Study to optimize runs of homozygosity (ROH) calling by maximizing the correlation between parental genetic relatedness and offspring ROHs. We then estimate the effect of maternal, paternal and fetal autozygosity and that of autozygosity mapping (common segments and gene burden test) on the timing of spontaneous onset of delivery. The correlation between offspring ROH using a variety of parameters and parental genetic relatedness ranged between −0.2 and 0.6, revealing the importance of the minimum number of genetic variants included in an ROH and the use of genetic distance. The optimized compared to predefined parameters showed a ≃45% higher correlation between parental genetic relatedness and offspring ROH. We found no evidence of an effect of maternal, paternal nor fetal overall autozygosity on spontaneous delivery timing. Yet, through autozygosity mapping, we identified three maternal loci TBC1D1, SIGLECs and EDN1 gene regions reducing the median time-to-spontaneous onset of delivery by ≃2–5% (P-value < 2.3 × 10⁻⁶). We also found suggestive evidence of a fetal locus at 3q22.2, near the RYK gene region (P-value = 2.0 × 10⁻⁶). Autozygosity mapping may provide new insights on the genetic determinants of delivery timing beyond traditional genome-wide association studies, but particular and rigorous attention should be given to ROH calling parameter selection.

A multiple coefficient of determination-based method for parsing SNPs that correlate with mRNA expression

In this study, we present a novel, multiple coefficient of determination (R²_M)-based method for parsing SNPs located within the chromosomal neighborhood of a gene into semi-independent families, each of which corresponds to one or more functional variants that regulate transcription of the gene. Specifically, our method utilizes a matrix equation framework to calculate R²_M values for SNPs within a chromosome region of interest (ROI) based upon the choices of 1-4 "index" SNPs (iSNPs) that serve as proxies for underlying regulatory variants. Exhaustive testing of sets of 1-4 candidate iSNPs identifies iSNP models that best account for estimated R² values derived from single-variable linear regression analysis of correlations between mRNA expression and genotypes of individual SNPs. Subsequent genotype-based estimation of pairwise r² linkage disequilibrium (LD) coefficients between each iSNP and the other ROI SNPs allows the SNPs to be parsed into semi-independent families. Analysis of mRNA expression and genotypes data downloaded from Gene Expression Omnibus (GEO) and database for Genotypes and Phenotypes (dbGAP) demonstrates the usefulness of this method for parsing SNPs based on experimental data. We believe that this method will be widely applicable for the analysis of the genetic basis of mRNA expression and visualizing the contributions of multiple genetic variants to the regulation of individual genes.

Crossing Phenotype Heritability and Candidate Gene Expression in Grafted Black-Lipped Pearl Oyster Pinctada margaritifera, an Animal Chimera

Grafting mantle tissue of a donor pearl oyster into the gonad of a recipient oyster results in the formation of a chimera, the pearl sac. The phenotypic variations of this chimera are hypothesized to be the result of interactions between the donor and recipient genomes. In this study, the heritability of phenotypic variation and its association with gene expression were investigated for the first time during Pinctada margaritifera pearl production. Genetic variance was evaluated at different levels, 1) before the graft operation (expression in graft tissue), 2) after grafting (pearl sac tissue expression in chimera), and 3) on the product of the graft (pearl phenotype traits) based on controlled biparental crosses and the F1 generation. Donor-related genetic parameter estimates clearly demonstrate heritability for nacre weight and thickness, darkness and color, and surface defects and grade, which signifies a genetic basis in the donor oyster. In graft relative gene expression, the value of heritability was superior to 0.20 in for almost all genes; whereas in pearl sac, heritability estimates were low (h2 < 0.10; except for CALC1 and Aspein). Pearl sac expression seems to be more influenced by residual variance than the graft, which can be explained by environmental effects that influence pearls sac gene expression and act as a recipient additive genetic component. The interactions between donor and recipient are very complex, and further research is required to understand the role of the recipient oysters on pearl phenotypic and gene expression variances.

Comprehensive Multiple eQTL Detection and Its Application to GWAS Interpretation

Expression QTL (eQTL) detection has emerged as an important tool for unraveling the relationship between genetic risk factors and disease or clinical phenotypes. Most studies are predicated on the assumption that only a single causal variant explains the association signal in each interval. This greatly simplifies the statistical modeling, but is liable to biases in scenarios where multiple local causal-variants are responsible. Here, our primary goal was to address the prevalence of secondary cis-eQTL signals regulating peripheral blood gene expression locally, utilizing two large human cohort studies, each >2500 samples with accompanying whole genome genotypes. The CAGE (Consortium for the Architecture of Gene Expression) dataset is a compendium of Illumina microarray studies, and the Framingham Heart Study is a two-generation Affymetrix dataset. We also describe Bayesian colocalization analysis of the extent of sharing of cis-eQTL detected in both studies as well as with the BIOS RNAseq dataset. Stepwise conditional modeling demonstrates that multiple eQTL signals are present for ∼40% of over 3500 eGenes in both microarray datasets, and that the number of loci with additional signals reduces by approximately two-thirds with each conditioning step. Although <20% of the peak signals across platforms fine map to the same credible interval, the colocalization analysis finds that as many as 50-60% of the primary eQTL are actually shared. Subsequently, colocalization of eQTL signals with GWAS hits detected 1349 genes whose expression in peripheral blood is associated with 591 human phenotype traits or diseases, including enrichment for genes with regulatory functions. At least 10%, and possibly as many as 40%, of eQTL-trait colocalized signals are due to nonprimary cis-eQTL peaks, but just one-quarter of these colocalization signals replicated across the gene expression datasets. Our results are provided as a web-based resource for visualization of multi-site regulation of gene expression and its association with human complex traits and disease states.

Genome-Wide Variants Shared Between Smoking Quantity and Schizophrenia on 15q25 Are Associated With CHRNA5 Expression in the Brain

Cigarette smokers with schizophrenia consume more cigarettes than smokers in the general population. Schizophrenia and smoking quantity may have shared genetic liability. Genome-wide association studies (GWASs) of schizophrenia and smoking quantity have highlighted a biological pleiotropy in which a robust 15q25 locus affects both traits. To identify the genetic variants shared between these traits on 15q25, we used summary statistics from large-scale GWAS meta-analyses of schizophrenia in the Psychiatric Genomics Consortium 2 and smoking quantity assessed by cigarettes smoked per day in the Tobacco and Genetics Consortium. To evaluate the regulatory potential of the shared genetic variants, expression quantitative trait loci analysis in 10 postmortem brain regions was performed using the BRAINEAC dataset in 134 neuropathologically normal individuals. Twenty-two genetic variants on 15q25 were associated with both smoking quantity and schizophrenia at the genome-wide significance level (P < 5.00 × 10-8). Major alleles of all variants were associated with higher smoking quantity and risk of schizophrenia. These genetic variants were associated with PSMA4, CHRNA3, and CHRNB4 expression in specific brain regions (lowest P = 4.81 × 10-4) and with CHRNA5 expression in multiple brain regions (lowest P = 8.70 × 10-6). Risk-associated major alleles of these variants were commonly associated with higher expression in several brain regions, excluding the medulla, at the transcript level. In addition, the risk-associated major allele at rs637137 was associated with higher CHRNA5 expression at the specific exon level in multiple brain regions (lowest P = 2.37 × 10-5). Our findings suggest that genome-wide variants shared between smoking quantity and schizophrenia contribute to a common pathophysiology underlying these traits involving altered CHRNA5 expression in the brain.

Apple fruit acidity is genetically diversified by natural variations in three hierarchical epistatic genes: MdSAUR37, MdPP2CH and MdALMTII

Many efforts have been made to map quantitative trait loci (QTLs) to facilitate practical marker-assisted selection (MAS) in plants. In the present study, using MapQTL and BSA-seq (bulk segregant analysis using next generation sequencing) with two independent pedigree-based populations, we identified four major genome-wide QTLs responsible for apple fruit acidity. Candidate genes were screened in major QTL regions, and three functional gene markers, including a non-synonymous A/G single-nucleotide polymorphism (SNP) in the coding region of MdPP2CH, a 36-bp insertion in the promoter of MdSAUR37 and a previously reported SNP in MdALMTII, were validated to influence the malate content of apple fruits. In addition, MdPP2CH inactivated three vacuolar H⁺ -ATPases (MdVHA-A3, MdVHA-B2 and MdVHA-D2) and one aluminium-activated malate transporter (MdALMTII) via dephosphorylation and negatively influenced fruit malate accumulation. The dephosphotase activity of MdPP2CH was suppressed by MdSAUR37, which implied a higher hierarchy of genetic interaction. Therefore, the MdSAUR37/MdPP2CH/MdALMTII chain cascaded hierarchical epistatic genetic effects to precisely determine apple fruit malate content. An A/G SNP (-1010) on the MdMYB44 promoter region from a major QTL (qtl08.1) was closely associated with fruit malate content. The predicted phenotype values (PPVs) were estimated using the tentative genotype values of the gene markers, and the PPVs were significantly correlated with the observed phenotype values. Our findings provide an insight into plant genome-based selection in apples and will aid in conducting research to understand the fundamental physiological basis of quantitative genetics.

Trans-eQTLs identified in whole blood have limited influence on complex disease biology

Trans-eQTLs have been implicated in complex traits and common diseases, but many were initially identified on the basis of having an effect in cis, and there has been no assessment of the significance of the overlap in relation to chance expectations. Here, we investigated whether trans-expression quantitative trait loci (eQTL) associations identified in whole blood contribute to variance in complex traits by determining (1) whether genome-wide significant (GWS) single-nucleotide polymorphisms (SNPs) were enriched for trans-eQTL (including trans-only eQTL), and (2) whether the genomic regions surrounding associated trans-genes were enriched for statistical associations in the relevant GWAS. On average for a given phenotype, we identify 4.8% of GWS SNPs overlapping with trans-eQTL present in blood, and show that for the majority of these phenotypes, this observation does not exceed that expected by chance. Likewise, we observe no enrichment for genetic associations with the GWAS phenotype in the regions surrounding the linked trans-genes, with the exception of rheumatoid arthritis. Interestingly, the GWS SNPs for each phenotype were consistently more enriched for unique trans-eQTL SNPs than trans-eQTL SNP-probe pairs (p = 4 × 10^-7), with schizophrenia the only exception. This relative enrichment for trans-eQTL SNPs over trans-eQTL SNP-probe pairs implies that trait-associated trans-eQTL SNPs in whole blood are less likely to be 'master regulators' than random trans-eQTL SNPs. Taken together, these results suggest little evidence for the role of blood-based trans-eQTL in complex traits and disease, although this may reflect the finite size of currently available data sets and our findings may not hold for trans-eQTLs in more trait-relevant tissues. All software is publically available at https://github.com/IMB-Computational-Genomics-Lab/eqtlOverlapper .

Estimation of cis-eQTL effect sizes using a log of linear model

The study of expression Quantitative Trait Loci (eQTL) is an important problem in genomics and biomedicine. While detection (testing) of eQTL associations has been widely studied, less work has been devoted to the estimation of eQTL effect size. To reduce false positives, detection methods frequently rely on linear modeling of rank-based normalized or log-transformed gene expression data. Unfortunately, these approaches do not correspond to the simplest model of eQTL action, and thus yield estimates of eQTL association that can be uninterpretable and inaccurate. In this article, we propose a new, log-of-linear model for eQTL action, termed ACME, that captures allelic contributions to cis-acting eQTLs in an additive fashion, yielding effect size estimates that correspond to a biologically coherent model of cis-eQTLs. We describe a non-linear least-squares algorithm to fit the model by maximum likelihood, and obtain corresponding p-values. We perform careful investigation of the model using a combination of simulated data and data from the Genotype Tissue Expression (GTEx) project. Our results reveal little evidence for dominance effects, a parsimonious result that accords with a simple biological model for allele-specific expression and supports use of the ACME model. We show that Type-I error is well-controlled under our approach in a realistic setting, so that rank-based normalizations are unnecessary. Furthermore, we show that such normalizations can be detrimental to power and estimation accuracy under the proposed model. We then show, through effect size analyses of whole-genome cis-eQTLs in the GTEx data, that using standard normalizations instead of ACME noticeably affects the ranking and sign of estimates.

Association testing of bisulfite-sequencing methylation data via a Laplace approximation

Motivation

Epigenome-wide association studies can provide novel insights into the regulation of genes involved in traits and diseases. The rapid emergence of bisulfite-sequencing technologies enables performing such genome-wide studies at the resolution of single nucleotides. However, analysis of data produced by bisulfite-sequencing poses statistical challenges owing to low and uneven sequencing depth, as well as the presence of confounding factors. The recently introduced Mixed model Association for Count data via data AUgmentation (MACAU) can address these challenges via a generalized linear mixed model when confounding can be encoded via a single variance component. However, MACAU cannot be used in the presence of multiple variance components. Additionally, MACAU uses a computationally expensive Markov Chain Monte Carlo (MCMC) procedure, which cannot directly approximate the model likelihood.

Results

We present a new method, Mixed model Association via a Laplace ApproXimation (MALAX), that is more computationally efficient than MACAU and allows to model multiple variance components. MALAX uses a Laplace approximation rather than MCMC based approximations, which enables to directly approximate the model likelihood. Through an extensive analysis of simulated and real data, we demonstrate that MALAX successfully addresses statistical challenges introduced by bisulfite-sequencing while controlling for complex sources of confounding, and can be over 50% faster than the state of the art.

Availability and Implementation

The full source code of MALAX is available at https://github.com/omerwe/MALAX.

Supplementary information

Supplementary data are available at Bioinformatics online.

Familial resemblances in human whole blood transcriptome

Background

Considering the implication of gene expression in the susceptibility of chronic diseases and the familial clustering of chronic diseases, the study of familial resemblances in gene expression levels is then highly relevant. Few studies have considered the contribution of both genetic and common environmental effects to familial resemblances in whole blood gene expression levels. The objective is to quantify the contribution of genetic and common environmental effects in the familial resemblances of whole blood genome-wide gene expression levels. We also make comparisons with familial resemblances in blood leukocytes genome-wide DNA methylation levels in the same cohort in order to further investigate biological mechanisms.

Results

Maximal heritability, genetic heritability, and common environmental effect were computed for all probes (20.6%, 15.6%, and 5.0% respectively) and for probes showing a significant familial effect (78.1%, 60.1%, and 18.0% respectively). Pairwise phenotypic correlations between gene expression and DNA methylation levels adjusted for blood cell heterogeneity were computed for probes showing significant familial effect. A total of 78 probe pairs among the 7,618,401 possible pairs passed Bonferroni correction (corrected P-value = 6.56 × 10^− 9). Significant genetic correlations between gene expression and DNA methylation levels were found for 25 probe pairs (absolute genetic correlation of 0.97).

Conclusions

Familial resemblances in gene expression levels were mainly attributable to genetic factors, but common environmental effect also played a role especially in probes showing a significant familial effect. Probes and CpG sites with familial effect seem to be under a strong shared genetic control.

Electronic supplementary material

The online version of this article (10.1186/s12864-018-4698-6) contains supplementary material, which is available to authorized users.

Endometriosis risk alleles at 1p36.12 act through inverse regulation of CDC42 and LINC00339

Genome-wide association studies (GWAS) have identified markers within the WNT4 region on chromosome 1p36.12 showing consistent and strong association with increasing endometriosis risk. Fine mapping using sequence and imputed genotype data has revealed strong candidates for the causal SNPs within these critical regions; however, the molecular pathogenesis of these SNPs is currently unknown. We used gene expression data collected from whole blood from 862 individuals and endometrial tissue from 136 individuals from independent populations of European descent to examine the mechanism underlying endometriosis susceptibility. Association mapping results from 7,090 individuals (2,594 cases and 4,496 controls) supported rs3820282 as the SNP with the strongest association for endometriosis risk (P = 1.84 × 10−5, OR = 1.244 (1.126-1.375)). SNP rs3820282 is a significant eQTL in whole blood decreasing expression of LINC00339 (also known as HSPC157) and increasing expression of CDC42 (P = 2.0 ×10−54 and 4.5x10−4 respectively). The largest effects were for two LINC00339 probes (P = 2.0 ×10−54; 1.0 × 10−34). The eQTL for LINC00339 was also observed in endometrial tissue (P = 2.4 ×10−8) with the same direction of effect for both whole blood and endometrial tissue. There was no evidence for eQTL effects for WNT4. Chromatin conformation capture provides evidence for risk SNPs interacting with the promoters of both LINC00339 and CDC4 and luciferase reporter assays suggest the risk SNP rs12038474 is located in a transcriptional silencer for CDC42 and the risk allele increases expression of CDC42. However, no effect of rs3820282 was observed in the LINC00339 expression in Ishikawa cells. Taken together, our results suggest that SNPs increasing endometriosis risk in this region act through CDC42, but further functional studies are required to rule out inverse regulation of both LINC00339 and CDC42.

An ancestry-based approach for detecting interactions

BACKGROUND

Epistasis and gene-environment interactions are known to contribute significantly to variation of complex phenotypes in model organisms. However, their identification in human association studies remains challenging for myriad reasons. In the case of epistatic interactions, the large number of potential interacting sets of genes presents computational, multiple hypothesis correction, and other statistical power issues. In the case of gene-environment interactions, the lack of consistently measured environmental covariates in most disease studies precludes searching for interactions and creates difficulties for replicating studies.

RESULTS

In this work, we develop a new statistical approach to address these issues that leverages genetic ancestry, defined as the proportion of ancestry derived from each ancestral population (e.g., the fraction of European/African ancestry in African Americans), in admixed populations. We applied our method to gene expression and methylation data from African American and Latino admixed individuals, respectively, identifying nine interactions that were significant at P<5×10-8. We show that two of the interactions in methylation data replicate, and the remaining six are significantly enriched for low P-values (P<1.8×10-6).

CONCLUSION

We show that genetic ancestry can be a useful proxy for unknown and unmeasured covariates in the search for interaction effects. These results have important implications for our understanding of the genetic architecture of complex traits.

Testing Two Evolutionary Theories of Human Aging with DNA Methylation Data

The evolutionary theories of mutation accumulation (MA) and disposable soma (DS) provide possible explanations for the existence of human aging. To better understand the relative importance of these theories, we devised a test to identify MA- and DS-consistent sites across the genome using familial DNA methylation data. Two key characteristics of DNA methylation allowed us to do so. First, DNA methylation exhibits distinct and widespread changes with age, with numerous age-differentially-methylated sites observed across the genome. Second, many sites show heritable DNA methylation patterns within families. We extended heritability predictions of MA and DS to DNA methylation, predicting that MA-consistent age-differentially-methylated sites will show increasing heritability with age, while DS-consistent sites will show the opposite. Variance components models were used to test for changing heritability of methylation with age at 48,601 age-differentially-methylated sites across the genome in 610 individuals from 176 families. Of these, 102 sites showed significant MA-consistent increases in heritability with age, while 2266 showed significant DS-consistent decreases in heritability. These results suggest that both MA and DS play a role in explaining aging and aging-related changes, and that while the majority of DNA methylation changes observed in aging are consistent with epigenetic drift, targeted changes exist and may mediate effects of aging-related genes.

Constraints on eQTL Fine Mapping in the Presence of Multisite Local Regulation of Gene Expression

Expression quantitative trait locus (eQTL) detection has emerged as an important tool for unraveling of the relationship between genetic risk factors and disease or clinical phenotypes. Most studies use single marker linear regression to discover primary signals, followed by sequential conditional modeling to detect secondary genetic variants affecting gene expression. However, this approach assumes that functional variants are sparsely distributed and that close linkage between them has little impact on estimation of their precise location and the magnitude of effects. We describe a series of simulation studies designed to evaluate the impact of linkage disequilibrium (LD) on the fine mapping of causal variants with typical eQTL effect sizes. In the presence of multisite regulation, even though between 80 and 90% of modeled eSNPs associate with normally distributed traits, up to 10% of all secondary signals could be statistical artifacts, and at least 5% but up to one-quarter of credible intervals of SNPs within r² > 0.8 of the peak may not even include a causal site. The Bayesian methods eCAVIAR and DAP (Deterministic Approximation of Posteriors) provide only modest improvement in resolution. Given the strong empirical evidence that gene expression is commonly regulated by more than one variant, we conclude that the fine mapping of causal variants needs to be adjusted for multisite influences, as conditional estimates can be highly biased by interference among linked sites, but ultimately experimental verification of individual effects is needed. Presumably similar conclusions apply not just to eQTL mapping, but to multisite influences on fine mapping of most types of quantitative trait.

Spatial and temporal expression patterns of genes around nine neuroticism-associated loci

Neuroticism is a high-order personality trait. Individuals with higher neuroticism have increased risks of various psychiatric disorders and physical health outcomes. Neuroticism is related to physiological differences in the brain. A recent genome-wide association study identified nine distinct genomic loci that contribute to neuroticism. Brain development and function depend on the precise regulation of gene expression, which is differentially regulated across brain regions and developmental stages. Using multiple publicly available human post-mortem databases, we investigated, in brain and non-brain tissues and across several developmental life stages, the spatial and temporal expression patterns of genes arising from nine neuroticism-associated loci. Functional gene-network analysis for neuroticism-associated genes was performed. The spatial expression analysis revealed that the nearest genes (GRIK3, SRP9, KLHL2, PTPRD, ELAVL2, CRHR1 and CELF4) from index single-nucleotide polymorphisms (SNPs) at the nine loci were intensively enriched in the brain compared with their representation in non-brain tissues (p<1.56×10^-3). The nearest genes associated with the glutamate receptor activity network consisted mainly of GRIK3 (FDR q=4.25×10^-2). The temporal expression analysis revealed that the neuroticism-associated genes were divided into three expression patterns: KLHL2, CELF4 and CRHR1 were preferentially expressed during postnatal stages; PTPRD, ELAVL2 and MFHAS1 were expressed during prenatal stages; and the other three genes were not expressed during specific life stages. These findings suggest that the glutamate network might be a target for investigating the neurobiological mechanisms underlying susceptibilities to higher neuroticism and several psychiatric disorders and that neuroticism is mediated by genes specifically expressed in the brain during several developmental stages.

Evidence for mitochondrial genetic control of autosomal gene expression

The mitochondrial and nuclear genomes coordinate and co-evolve in eukaryotes in order to adapt to environmental changes. Variation in the mitochondrial genome is capable of affecting expression of genes on the nuclear genome. Sex-specific mitochondrial genetic control of gene expression has been demonstrated in Drosophila melanogaster, where males were found to drive most of the total variation in gene expression. This has potential implications for male-related health and disease resulting from variation in mtDNA solely inherited from the mother. We used a family-based study comprised of 47,323 gene expression probes and 78 mitochondrial SNPs (mtSNPs) from n = 846 individuals to examine the extent of mitochondrial genetic control of gene expression in humans. This identified 15 significant probe-mtSNP associations (P<10-8) corresponding to 5 unique genes on the mitochondrial and nuclear genomes, with three of these genes corresponding to mitochondrial genetic control of gene expression in the nuclear genome. The associated mtSNPs for three genes (one cis and two trans associations) were replicated (P < 0.05) in an independent dataset of n = 452 unrelated individuals. There was no evidence for sexual dimorphic gene expression in any of these five probes. Sex-specific effects were examined by applying our analysis to males and females separately and testing for differences in effect size. The MEST gene was identified as having the most significantly different effect sizes across the sexes (P≈10-7). MEST was similarly expressed in males and females with the G allele; however, males with the C allele are highly expressed for MEST, while females show no expression of the gene. This study provides evidence for the mitochondrial genetic control of expression of several genes in humans, with little evidence found for sex-specific effects.

Gene co-expression network connectivity is an important determinant of selective constraint

While several studies have investigated general properties of the genetic architecture of natural variation in gene expression, few of these have considered natural, outbreeding populations. In parallel, systems biology has established that a general feature of biological networks is that they are scale-free, rendering them buffered against random mutations. To date, few studies have attempted to examine the relationship between the selective processes acting to maintain natural variation of gene expression and the associated co-expression network structure. Here we utilised RNA-Sequencing to assay gene expression in winter buds undergoing bud flush in a natural population of Populus tremula, an outbreeding forest tree species. We performed expression Quantitative Trait Locus (eQTL) mapping and identified 164,290 significant eQTLs associating 6,241 unique genes (eGenes) with 147,419 unique SNPs (eSNPs). We found approximately four times as many local as distant eQTLs, with local eQTLs having significantly higher effect sizes. eQTLs were primarily located in regulatory regions of genes (UTRs or flanking regions), regardless of whether they were local or distant. We used the gene expression data to infer a co-expression network and investigated the relationship between network topology, the genetic architecture of gene expression and signatures of selection. Within the co-expression network, eGenes were underrepresented in network module cores (hubs) and overrepresented in the periphery of the network, with a negative correlation between eQTL effect size and network connectivity. We additionally found that module core genes have experienced stronger selective constraint on coding and non-coding sequence, with connectivity associated with signatures of selection. Our integrated genetics and genomics results suggest that purifying selection is the primary mechanism underlying the genetic architecture of natural variation in gene expression assayed in flushing leaf buds of P. tremula and that connectivity within the co-expression network is linked to the strength of purifying selection.

The Genetic Architecture of Gene Expression in Peripheral Blood

We analyzed the mRNA levels for 36,778 transcript expression traits (probes) from 2,765 individuals to comprehensively investigate the genetic architecture and degree of missing heritability for gene expression in peripheral blood. We identified 11,204 cis and 3,791 trans independent expression quantitative trait loci (eQTL) by using linear mixed models to perform genome-wide association analyses. Furthermore, using information on both closely and distantly related individuals, heritability was estimated for all expression traits. Of the set of expressed probes (15,966), 10,580 (66%) had an estimated narrow-sense heritability (h²) greater than zero with a mean (median) value of 0.192 (0.142). Across these probes, on average the proportion of genetic variance explained by all eQTL (h_COJO²) was 31% (0.060/0.192), meaning that 69% is missing, with the sentinel SNP of the largest eQTL explaining 87% (0.052/0.060) of the variance attributed to all identified cis- and trans-eQTL. For the same set of probes, the genetic variance attributed to genome-wide common (MAF > 0.01) HapMap 3 SNPs (h_g²) accounted for on average 48% (0.093/0.192) of h². Taken together, the evidence suggests that approximately half the genetic variance for gene expression is not tagged by common SNPs, and of the variance that is tagged by common SNPs, a large proportion can be attributed to identifiable eQTL of large effect, typically in cis. Finally, we present evidence that, compared with a meta-analysis, using individual-level data results in an increase of approximately 50% in power to detect eQTL.

Integrated genome-wide analysis of expression quantitative trait loci aids interpretation of genomic association studies

Background

Identification of single nucleotide polymorphisms (SNPs) associated with gene expression levels, known as expression quantitative trait loci (eQTLs), may improve understanding of the functional role of phenotype-associated SNPs in genome-wide association studies (GWAS). The small sample sizes of some previous eQTL studies have limited their statistical power. We conducted an eQTL investigation of microarray-based gene and exon expression levels in whole blood in a cohort of 5257 individuals, exceeding the single cohort size of previous studies by more than a factor of 2.

Results

We detected over 19,000 independent lead cis-eQTLs and over 6000 independent lead trans-eQTLs, targeting over 10,000 gene targets (eGenes), with a false discovery rate (FDR) < 5%. Of previously published significant GWAS SNPs, 48% are identified to be significant eQTLs in our study. Some trans-eQTLs point toward novel mechanistic explanations for the association of the SNP with the GWAS-related phenotype. We also identify 59 distinct blocks or clusters of trans-eQTLs, each targeting the expression of sets of six to 229 distinct trans-eGenes. Ten of these sets of target genes are significantly enriched for microRNA targets (FDR < 5%). Many of these clusters are associated in GWAS with multiple phenotypes.

Conclusions

These findings provide insights into the molecular regulatory patterns involved in human physiology and pathophysiology. We illustrate the value of our eQTL database in the context of a recent GWAS meta-analysis of coronary artery disease and provide a list of targeted eGenes for 21 of 58 GWAS loci.

Electronic supplementary material

The online version of this article (doi:10.1186/s13059-016-1142-6) contains supplementary material, which is available to authorized users.

Genome-wide analysis of blood gene expression in migraine implicates immune-inflammatory pathways

Background Typical migraine is a frequent, debilitating and painful headache disorder with an estimated heritability of about 50%. Although genome-wide association (GWA) studies have identified over 40 single nucleotide polymorphisms associated with migraine, further research is required to determine their biological role in migraine susceptibility. Therefore, we performed a study of genome-wide gene expression in a large sample of 83 migraine cases and 83 non-migraine controls to determine whether altered expression levels of genes and pathways could provide insights into the biological mechanisms underlying migraine. Methods We assessed whole blood gene expression data for 17994 expression probes measured using IlluminaHT-12 v4.0 BeadChips. Differential expression was assessed using multivariable logistic regression. Gene expression probes with a nominal p value < 0.05 were classified as differentially expressed. We identified modules of co-regulated genes and tested them for enrichment of differentially expressed genes and functional pathways using a false discovery rate <0.05. Results Association analyses between migraine and probe expression levels, adjusted for age and gender, revealed an excess of small p values, but there was no significant single-probe association after correction for multiple testing. Network analysis of pooled expression data identified 10 modules of co-expressed genes. One module harboured a significant number of differentially expressed genes and was strongly enriched with immune-inflammatory pathways, including multiple pathways expressed in microglial cells. Conclusions These data suggest immune-inflammatory pathways play an important role in the pathogenesis, manifestation, and/or progression of migraine in some patients. Furthermore, gene-expression associations are measurable in whole blood, suggesting the analysis of blood gene expression can inform our understanding of the biological mechanisms underlying migraine, identify biomarkers, and facilitate the discovery of novel pathways and thus determine new targets for drug therapy.

Regulatory Architecture of Gene Expression Variation in the Threespine Stickleback Gasterosteus aculeatus

Much adaptive evolutionary change is underlain by mutational variation in regions of the genome that regulate gene expression rather than in the coding regions of the genes themselves. An understanding of the role of gene expression variation in facilitating local adaptation will be aided by an understanding of underlying regulatory networks. Here, we characterize the genetic architecture of gene expression variation in the threespine stickleback (Gasterosteus aculeatus), an important model in the study of adaptive evolution. We collected transcriptomic and genomic data from 60 half-sib families using an expression microarray and genotyping-by-sequencing, and located expression quantitative trait loci (eQTL) underlying the variation in gene expression in liver tissue using an interval mapping approach. We identified eQTL for several thousand expression traits. Expression was influenced by polymorphism in both cis- and trans-regulatory regions. Trans-eQTL clustered into hotspots. We did not identify master transcriptional regulators in hotspot locations: rather, the presence of hotspots may be driven by complex interactions between multiple transcription factors. One observed hotspot colocated with a QTL recently found to underlie salinity tolerance in the threespine stickleback. However, most other observed hotspots did not colocate with regions of the genome known to be involved in adaptive divergence between marine and freshwater habitats.

The autoimmune risk gene ZMIZ1 is a vitamin D responsive marker of a molecular phenotype of multiple sclerosis

Multiple Sclerosis (MS) is a neurological condition driven in part by immune cells from the peripheral circulation, the targets for current successful therapies. The autoimmune and MS risk gene ZMIZ1 is underexpressed in blood in people with MS. We show that, from three independent sets of transcriptomic data, expression of ZMIZ1 is tightly correlated with that of hundreds of other genes. Further we show expression is partially heritable (heritability 0.26), relatively stable over time, predominantly in plasmacytoid dendritic cells and non-classical monocytes, and that levels of ZMIZ1 protein expression are reduced in MS. ZMIZ1 gene expression is increased in response to calcipotriol (1,25 Vitamin D3) (p < 0.0003) and associated with Epstein Barr Virus (EBV) EBNA-1 antibody titre (p < 0.004). MS therapies fingolimod and dimethyl fumarate altered blood ZMIZ1 gene expression compared to untreated MS. The phenotype indicates susceptibility to MS, and may correspond with clinical response and represent a novel clinical target.

Variability of 128 schizophrenia-associated gene variants across distinct ethnic populations

Schizophrenia is a common polygenetic disease affecting 0.5-1% of individuals across distinct ethnic populations. PGC-II, the largest genome-wide association study investigating genetic risk factors for schizophrenia, previously identified 128 independent schizophrenia-associated genetic variants (GVs). The current study examined the genetic variability of GVs across ethnic populations. To assess the genetic variability across populations, the 'variability indices' (VIs) of the 128 schizophrenia-associated GVs were calculated. We used 2504 genomes from the 1000 Genomes Project taken from 26 worldwide healthy samples comprising five major ethnicities: East Asian (EAS: n=504), European (EUR: n=503), African (AFR: n=661), American (AMR: n=347) and South Asian (SAS: n=489). The GV with the lowest variability was rs36068923 (VI=1.07). The minor allele frequencies (MAFs) were 0.189, 0.192, 0.256, 0.183 and 0.194 for EAS, EUR, AFR, AMR and SAS, respectively. The GV with the highest variability was rs7432375 (VI=9.46). The MAFs were 0.791, 0.435, 0.041, 0.594 and 0.508 for EAS, EUR, AFR, AMR and SAS, respectively. When we focused on the EAS and EUR population, the allele frequencies of 86 GVs significantly differed between the EAS and EUR (P<3.91 × 10^-4). The GV with the highest variability was rs4330281 (P=1.55 × 10^-138). The MAFs were 0.023 and 0.519 for the EAS and EUR, respectively. The GV with the lowest variability was rs2332700 (P=9.80 × 10^-1). The MAFs were similar between these populations (that is, 0.246 and 0.247 for the EAS and EUR, respectively). Interestingly, the mean allele frequencies of the GVs did not significantly differ between these populations (P>0.05). Although genetic heterogeneities were observed in the schizophrenia-associated GVs across ethnic groups, the combination of these GVs might increase the risk of schizophrenia.

Systems Genetics as a Tool to Identify Master Genetic Regulators in Complex Disease

Systems genetics stems from systems biology and similarly employs integrative modeling approaches to describe the perturbations and phenotypic effects observed in a complex system. However, in the case of systems genetics the main source of perturbation is naturally occurring genetic variation, which can be analyzed at the systems-level to explain the observed variation in phenotypic traits. In contrast with conventional single-variant association approaches, the success of systems genetics has been in the identification of gene networks and molecular pathways that underlie complex disease. In addition, systems genetics has proven useful in the discovery of master trans-acting genetic regulators of functional networks and pathways, which in many cases revealed unexpected gene targets for disease. Here we detail the central components of a fully integrated systems genetics approach to complex disease, starting from assessment of genetic and gene expression variation, linking DNA sequence variation to mRNA (expression QTL mapping), gene regulatory network analysis and mapping the genetic control of regulatory networks. By summarizing a few illustrative (and successful) examples, we highlight how different data-modeling strategies can be effectively integrated in a systems genetics study.

Autosomal genetic control of human gene expression does not differ across the sexes

BACKGROUND

Despite their nearly identical genomes, males and females differ in risk, incidence, prevalence, severity and age-at-onset of many diseases. Sexual dimorphism is also seen in human autosomal gene expression, and has largely been explored by examining the contribution of genotype-by-sex interactions to variation in gene expression.

RESULTS

In this study, we use data from a mixture of pedigree and unrelated individuals with verified European ancestry to investigate the sex-specific genetic architecture of gene expression measured in whole blood across n=1048 males and n=1005 females by treating gene expression intensities in the sexes as two distinct traits and estimating the genetic correlation (r _G) between them. These correlations measure the similarity of the combined additive genetic effects of all single-nucleotide polymorphisms across the autosomal chromosomes, and thus the level of common genetic control of gene expression across the sexes. Genetic correlations are estimated across the sexes for the expression levels of 12,528 autosomal gene expression probes using bivariate GREML, and tested for differences in autosomal genetic control of gene expression across the sexes. Overall, no deviation of the distribution of test statistics is observed from that expected under the null hypothesis of a common autosomal genetic architecture for gene expression across the sexes.

CONCLUSIONS

These results suggest that males and females share the same common genetic control of gene expression.

Local Adaptation of Sun-Exposure-Dependent Gene Expression Regulation in Human Skin

Sun-exposure is a key environmental variable in the study of human evolution. Several skin-pigmentation genes serve as classical examples of positive selection, suggesting that sun-exposure has significantly shaped worldwide genomic variation. Here we investigate the interaction between genetic variation and sun-exposure, and how this impacts gene expression regulation. Using RNA-Seq data from 607 human skin samples, we identified thousands of transcripts that are differentially expressed between sun-exposed skin and non-sun-exposed skin. We then tested whether genetic variants may influence each individual’s gene expression response to sun-exposure. Our analysis revealed 10 sun-exposure-dependent gene expression quantitative trait loci (se-eQTLs), including genes involved in skin pigmentation (SLC45A2) and epidermal differentiation (RASSF9). The allele frequencies of the RASSF9 se-eQTL across diverse populations correlate with the magnitude of solar radiation experienced by these populations, suggesting local adaptation to varying levels of sunlight. These results provide the first examples of sun-exposure-dependent regulatory variation and suggest that this variation has contributed to recent human adaptation.

Varying levels of sun-exposure across the world have significantly shaped human evolution. Previous analyses have found several skin pigmentation genes with evidence of strong evolutionary pressures throughout human history, manifesting as large differences in the frequency of genomic variants across populations. But even within populations, individuals respond differently to sun-exposure, suggesting variation in addition to the major differences in skin pigmentation across populations. Here we investigated whether genetic variants associate with response to sun-exposure within Europeans. To measure the response we analyzed gene expression in sun-exposed and non-sun-exposed skin, and identified ten genetic variants that associated with the sun-exposure response of nearby genes. One of these genetic variants, which associated with the sun-exposure response of the gene RASSF9, showed evidence of adaptation in humans in response to solar radiation. Together this evidence suggests that the regulation of gene expression is influenced by sun-exposure and that the sun-exposure dependent effect on RASSF9 expression may have had an effect on human fitness. To our knowledge, this is the first example of an environment-dependent regulatory variant with evidence of adaptation in humans.

Multikernel linear mixed models for complex phenotype prediction

Linear mixed models (LMMs) and their extensions have recently become the method of choice in phenotype prediction for complex traits. However, LMM use to date has typically been limited by assuming simple genetic architectures. Here, we present multikernel linear mixed model (MKLMM), a predictive modeling framework that extends the standard LMM using multiple-kernel machine learning approaches. MKLMM can model genetic interactions and is particularly suitable for modeling complex local interactions between nearby variants. We additionally present MKLMM-Adapt, which automatically infers interaction types across multiple genomic regions. In an analysis of eight case-control data sets from the Wellcome Trust Case Control Consortium and more than a hundred mouse phenotypes, MKLMM-Adapt consistently outperforms competing methods in phenotype prediction. MKLMM is as computationally efficient as standard LMMs and does not require storage of genotypes, thus achieving state-of-the-art predictive power without compromising computational feasibility or genomic privacy.

Shared genetic control of expression and methylation in peripheral blood

BACKGROUND

Expression QTLs and epigenetic marks are often employed to provide an insight into the possible biological mechanisms behind GWAS hits. A substantial proportion of the variation in gene expression and DNA methylation is known to be under genetic control. We address the proportion of genetic control that is shared between these two genomic features.

RESULTS

An exhaustive search for pairwise phenotypic correlations between gene expression and DNA methylation in samples from human blood (n = 610) was performed. Of the 5 × 10(9) possible pairwise tests, 0.36 % passed Bonferroni corrected p-value cutoff of 9.9 × 10(-12). We determined that the correlation structure between probe pairs was largely due to blood cell type specificity of the expression and methylation probes. Upon adjustment of the expression and methylation values for observed blood cellular composition (n = 422), the number of probe pairs which survived Bonferroni correction reduced by more than 5400 fold. Of the 614 correlated probe pairs located on the same chromosome, 75 % share at least one methylation and expression QTL at nominal 10(-5) p-value cutoff. Those probe pairs are located within 1Mbp window from each other and have a mean of absolute value of genetic correlation equal to 0.69, further demonstrating the high degree of shared genetic control.

CONCLUSIONS

Overall, this study demonstrates notable genetic covariance between DNA methylation and gene expression and reaffirms the importance of correcting for cell-counts in studies on non-homogeneous tissues.

Endometrial vezatin and its association with endometriosis risk

STUDY QUESTION

Do endometriosis risk-associated single nucleotide polymorphisms (SNPs) found at the 12q22 locus have effects on vezatin ( ITALIC! VEZT) expression?

SUMMARY ANSWER

The original genome-wide association study (GWAS) SNP (rs10859871), and other newly identified association signals, demonstrate strong evidence for ITALIC! cis-expression quantitative trait loci (eQTL) effects on ITALIC! VEZT expression.

WHAT IS KNOWN ALREADY

GWAS have identified several disease-risk loci (SNPs) associated with endometriosis. The SNP rs10859871 is located within the ITALIC! VEZT gene. ITALIC! VEZT expression is altered in the endometrium of endometriosis patients and is an excellent candidate for having a causal role in endometriosis. Most of the SNPs identified from GWAS are not located within the coding region of the genome. However, they are likely to have an effect on the regulation of gene expression. Genetic variants that affect levels of gene expression are called expression quantitative trait loci (eQTL).

STUDY DESIGN, SIZE, DURATION

Samples for genotyping and ITALIC! VEZT variant screening were drawn from women recruited for genetic studies in Australia/New Zealand and women undergoing surgery in a tertiary care centre. Coding variants for ITALIC! VEZT were screened in blood from 100 unrelated individuals (endometriosis-dense families) from the QIMR Berghofer Medical Research Institute dataset. SNPs at the 12q22 locus were imputed and reanalysed for their association with endometriosis. Reanalysis of endometriosis risk-association was performed on a final combined Australian dataset of 2594 cases and 4496 controls. Gene expression was performed on 136 endometrial samples. eQTL analysis in whole blood was performed on 862 individuals from the Brisbane Systems Genetics Study. Endometrial tissue-specific eQTL analysis was performed on 122 samples (eutopic endometrium) collected following laparoscopic surgery. VEZT protein expression studies employed ITALIC! n = 56 (western blotting) and ITALIC! n = 42 (immunohistochemistry) endometrial samples.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The women recruited for this study provided blood and/or endometrial tissue samples in a hospital setting. Genomic DNA was screened for common and coding variants. SNPs of interest in the 12q22 region were genotyped using Agena MassARRAY technology or Taqman SNP genotyping assay. Gene expression profiles from RNA extracted from blood and endometrial tissue samples were generated using Illumina whole-genome expression chips (Human HT-12 v4.0). Whole protein extracted from endometrium was used for VEZT western blots, and paraffin sections of endometrium were employed for VEZT immunohistochemistry semi-quantitative analysis.

MAIN RESULTS AND THE ROLE OF CHANCE

A total of 11 coding variants of ITALIC! VEZT (including one novel variant) were identified from an endometriosis-dense cohort. Polymorphic coding and imputed SNPs were combined with previous GWAS data to reanalyse the endometriosis risk association of the 12q22 region. The disease association signal at 12q22 was due to coding variants in ITALIC! VEZT or ITALIC! FGD6 (FYVE, RhoGEF and PH domain-containing 6) and SNPs with the strongest signals were either intronic or intergenic. We found strong evidence for ITALIC! VEZT cis-eQTLs with the sentinel SNP (rs10859871) in blood and endometrium, where the endometriosis risk allele (C) was associated with an increase in ITALIC! VEZT expression. We could not demonstrate this genotype-specific effect on VEZT protein expression in endometrium. However, we did observe a menstrual cycle stage specific increase in VEZT protein expression in endometrial glands, specific to the secretory phase ( ITALIC! P = 2.0 × 10(-4)).

LIMITATIONS, REASONS FOR CAUTION

In comparison to the blood sample datasets, the study numbers of endometrial tissues were substantially reduced. Protein studies failed to complement RNA results, also likely a reflection of the low study numbers in these experiments. ITALIC! In silico prediction tools used in this investigation are typically based on cell lines different to our tissues of interest, thus any functional annotations drawn from these approaches should be considered carefully. Therefore, functional studies on VEZT and related pathway components are still warranted to unequivocally implicate a causal role for VEZT in endometriosis pathophysiology.

WIDER IMPLICATIONS OF THE FINDINGS

GWAS have proven to be very valuable tools for deciphering complex diseases. Endometriosis is a text-book example of a complex disease, involving genetic, lifestyle and environmental influences. Our focused investigation of the 12q22 region validates an association with increased endometriosis risk. Endometriosis risk SNPs (including rs10859871) located within this locus demonstrated evidence for ITALIC! cis-eQTLs on ITALIC! VEZT expression. By examining women who possess an enhanced genetic risk of developing endometriosis, we have identified an effect on ITALIC! VEZT expression and therefore a potential gene/gene pathway in endometriosis disease establishment and development.

STUDY FUNDING/COMPETING INTERESTS

Funding for this work was provided by NHMRC Project Grants GNT1012245, GNT1026033, GNT1049472 and GNT1046880. G.W.M. is supported by the NHMRC Fellowship scheme (GNT1078399). S.J.H.-C. is supported by the J.N. Peters Bequest Fellowship. The authors declare no competing interests.

TRIAL REGISTRATION NUMBER

N/A.

Blood gene expression studies in migraine: Potential and caveats

BACKGROUND

Global gene expression analysis may be used to obtain insights into the functional processes underlying migraine. However, there is a shortage of high-quality post-mortem brain tissue samples for genetic analysis. One approach is to use a more accessible tissue as a surrogate, such as peripheral blood.

PURPOSE

Discuss the benefits and caveats of blood genomic profiling in migraine and its potential application in the development of biomarkers of migraine susceptibility and outcome. Demonstrate the utility of blood-based expression profiles in migraine by analysing pilot Illumina HT-12 expression data from 76 (38 case, 38 control) whole-blood samples.

CONCLUSION

Current evidence suggests peripheral blood is a biologically valid substrate for genetic studies of migraine, and may be used to identify biomarkers and therapeutic pathways. Pilot blood gene expression data confirm that expression profiles significantly differ between migraine case and non-migraine control individuals.

Weak Epistasis Generally Stabilizes Phenotypes in a Mouse Intercross

The extent and strength of epistasis is commonly unresolved in genetic studies, and observed epistasis is often difficult to interpret in terms of biological consequences or overall genetic architecture. We investigated the prevalence and consequences of epistasis by analyzing four body composition phenotypes—body weight, body fat percentage, femoral density, and femoral circumference—in a large F2 intercross of B6-lit/lit and C3.B6-lit/lit mice. We used Combined Analysis of Pleiotropy and Epistasis (CAPE) to examine interactions for the four phenotypes simultaneously, which revealed an extensive directed network of genetic loci interacting with each other, circulating IGF1, and sex to influence these phenotypes. The majority of epistatic interactions had small effects relative to additive effects of individual loci, and tended to stabilize phenotypes towards the mean of the population rather than extremes. Interactive effects of two alleles inherited from one parental strain commonly resulted in phenotypes closer to the population mean than the additive effects from the two loci, and often much closer to the mean than either single-locus model. Alternatively, combinations of alleles inherited from different parent strains contribute to more extreme phenotypes not observed in either parental strain. This class of phenotype-stabilizing interactions has effects that are close to additive and are thus difficult to detect except in very large intercrosses. Nevertheless, we found these interactions to be useful in generating hypotheses for functional relationships between genetic loci. Our findings suggest that while epistasis is often weak and unlikely to account for a large proportion of heritable variance, even small-effect genetic interactions can facilitate hypotheses of underlying biology in well-powered studies.

The role of statistical epistasis in the genetic architecture of complex traits has been of great interest to the genetics community since Fisher introduced the concept in 1918. However, assessing epistasis in human and model organism populations has been impeded by limited statistical power. To mitigate this limitation, we analyzed bone and body composition traits in an unusually large mouse intercross population of over 2000 mice, paired with a recently-developed computational approach that leverages information to detect interactions across multiple phenotypes. We discovered a large network of highly significant genetic interactions between variants that influence complex body composition traits. Although epistasis was abundant, the interaction network was dominated by epistasis that stabilizes phenotypes by reducing phenotypic deviation from the parent strains. Nevertheless, the observed network provides an overview of genetic architecture and specific hypotheses of how QTL combine to affect phenotypes. These findings suggest that epistatic effects are generally of lesser magnitude than main QTL effects, and therefore are unlikely to account for major components of variance, but also reinforce genetic interaction analysis as a potent tool for dissecting the biology of complex traits.

The low EOMES/TBX21 molecular phenotype in multiple sclerosis reflects CD56+ cell dysregulation and is affected by immunomodulatory therapies

Multiple Sclerosis (MS) is an autoimmune disease treated by therapies targeting peripheral blood cells. We previously identified that expression of two MS-risk genes, the transcription factors EOMES and TBX21 (ET), was low in blood from MS and stable over time. Here we replicated the low ET expression in a new MS cohort (p<0.0007 for EOMES, p<0.028 for TBX21) and demonstrate longitudinal stability (p<10(-4)) and high heritability (h(2)=0.48 for EOMES) for this molecular phenotype. Genes whose expression correlated with ET, especially those controlling cell migration, further defined the phenotype. CD56+ cells and other subsets expressed lower levels of Eomes or T-bet protein and/or were under-represented in MS. EOMES and TBX21 risk SNP genotypes, and serum EBNA-1 titres were not correlated with ET expression, but HLA-DRB1*1501 genotype was. ET expression was normalised to healthy control levels with natalizumab, and was highly variable for glatiramer acetate, fingolimod, interferon-beta, dimethyl fumarate.

Directional dominance on stature and cognition in diverse human populations

Homozygosity has long been associated with rare, often devastating, Mendelian disorders, and Darwin was one of the first to recognize that inbreeding reduces evolutionary fitness. However, the effect of the more distant parental relatedness that is common in modern human populations is less well understood. Genomic data now allow us to investigate the effects of homozygosity on traits of public health importance by observing contiguous homozygous segments (runs of homozygosity), which are inferred to be homozygous along their complete length. Given the low levels of genome-wide homozygosity prevalent in most human populations, information is required on very large numbers of people to provide sufficient power. Here we use runs of homozygosity to study 16 health-related quantitative traits in 354,224 individuals from 102 cohorts, and find statistically significant associations between summed runs of homozygosity and four complex traits: height, forced expiratory lung volume in one second, general cognitive ability and educational attainment (P < 1 × 10(-300), 2.1 × 10(-6), 2.5 × 10(-10) and 1.8 × 10(-10), respectively). In each case, increased homozygosity was associated with decreased trait value, equivalent to the offspring of first cousins being 1.2 cm shorter and having 10 months' less education. Similar effect sizes were found across four continental groups and populations with different degrees of genome-wide homozygosity, providing evidence that homozygosity, rather than confounding, directly contributes to phenotypic variance. Contrary to earlier reports in substantially smaller samples, no evidence was seen of an influence of genome-wide homozygosity on blood pressure and low density lipoprotein cholesterol, or ten other cardio-metabolic traits. Since directional dominance is predicted for traits under directional evolutionary selection, this study provides evidence that increased stature and cognitive function have been positively selected in human evolution, whereas many important risk factors for late-onset complex diseases may not have been.

Nonadditive Effects of Genes in Human Metabolomics

Genome-wide association studies (GWAS) are widely applied to analyze the genetic effects on phenotypes. With the availability of high-throughput technologies for metabolite measurements, GWAS successfully identified loci that affect metabolite concentrations and underlying pathways. In most GWAS, the effect of each SNP on the phenotype is assumed to be additive. Other genetic models such as recessive, dominant, or overdominant were considered only by very few studies. In contrast to this, there are theories that emphasize the relevance of nonadditive effects as a consequence of physiologic mechanisms. This might be especially important for metabolites because these intermediate phenotypes are closer to the underlying pathways than other traits or diseases. In this study we analyzed systematically nonadditive effects on a large panel of serum metabolites and all possible ratios (22,801 total) in a population-based study [Cooperative Health Research in the Region of Augsburg (KORA) F4, N = 1,785]. We applied four different 1-degree-of-freedom (1-df) tests corresponding to an additive, dominant, recessive, and overdominant trait model as well as a genotypic model with two degree-of-freedom (2-df) that allows a more general consideration of genetic effects. Twenty-three loci were found to be genome-wide significantly associated (Bonferroni corrected P ≤ 2.19 × 10(-12)) with at least one metabolite or ratio. For five of them, we show the evidence of nonadditive effects. We replicated 17 loci, including 3 loci with nonadditive effects, in an independent study (TwinsUK, N = 846). In conclusion, we found that most genetic effects on metabolite concentrations and ratios were indeed additive, which verifies the practice of using the additive model for analyzing SNP effects on metabolites.

Expression quantitative trait locus analysis for translational medicine

Expression quantitative trait locus analysis has emerged as an important component of efforts to understand how genetic polymorphisms influence disease risk and is poised to make contributions to translational medicine. Here we review how expression quantitative trait locus analysis is aiding the identification of which gene(s) within regions of association are causal for a disease or phenotypic trait; the narrowing down of the cell types or regulators involved in the etiology of disease; the characterization of drivers and modifiers of cancer; and our understanding of how different environments and cellular contexts can modify gene expression. We also introduce the concept of transcriptional risk scores as a means of refining estimates of individual liability to disease based on targeted profiling of the transcripts that are regulated by polymorphisms jointly associated with disease and gene expression.

Seasonal effects on gene expression

Many health conditions, ranging from psychiatric disorders to cardiovascular disease, display notable seasonal variation in severity and onset. In order to understand the molecular processes underlying this phenomenon, we have examined seasonal variation in the transcriptome of 606 healthy individuals. We show that 74 transcripts associated with a 12-month seasonal cycle were enriched for processes involved in DNA repair and binding. An additional 94 transcripts demonstrated significant seasonal variability that was largely influenced by blood cell count levels. These transcripts were enriched for immune function, protein production, and specific cellular markers for lymphocytes. Accordingly, cell counts for erythrocytes, platelets, neutrophils, monocytes, and CD19 cells demonstrated significant association with a 12-month seasonal cycle. These results demonstrate that seasonal variation is an important environmental regulator of gene expression and blood cell composition. Notable changes in leukocyte counts and genes involved in immune function indicate that immune cell physiology varies throughout the year in healthy individuals.

Dissecting the genetics of the human transcriptome identifies novel trait-related trans-eQTLs and corroborates the regulatory relevance of non-protein coding loci†

Genetics of gene expression (eQTLs or expression QTLs) has proved an indispensable tool for understanding biological pathways and pathomechanisms of trait-associated SNPs. However, power of most genome-wide eQTL studies is still limited. We performed a large eQTL study in peripheral blood mononuclear cells of 2112 individuals increasing the power to detect trans-effects genome-wide. Going beyond univariate SNP-transcript associations, we analyse relations of eQTLs to biological pathways, polygenetic effects of expression regulation, trans-clusters and enrichment of co-localized functional elements. We found eQTLs for about 85% of analysed genes, and 18% of genes were trans-regulated. Local eSNPs were enriched up to a distance of 5 Mb to the transcript challenging typically implemented ranges of cis-regulations. Pathway enrichment within regulated genes of GWAS-related eSNPs supported functional relevance of identified eQTLs. We demonstrate that nearest genes of GWAS-SNPs might frequently be misleading functional candidates. We identified novel trans-clusters of potential functional relevance for GWAS-SNPs of several phenotypes including obesity-related traits, HDL-cholesterol levels and haematological phenotypes. We used chromatin immunoprecipitation data for demonstrating biological effects. Yet, we show for strongly heritable transcripts that still little trans-chromosomal heritability is explained by all identified trans-eSNPs; however, our data suggest that most cis-heritability of these transcripts seems explained. Dissection of co-localized functional elements indicated a prominent role of SNPs in loci of pseudogenes and non-coding RNAs for the regulation of coding genes. In summary, our study substantially increases the catalogue of human eQTLs and improves our understanding of the complex genetic regulation of gene expression, pathways and disease-related processes.

DEGS2 polymorphism associated with cognition in schizophrenia is associated with gene expression in brain

A genome-wide association study of cognitive deficits in patients with schizophrenia in Japan found association with a missense genetic variant (rs7157599, Asn8Ser) in the delta(4)-desaturase, sphingolipid 2 (DEGS2) gene. A replication analysis using Caucasian samples showed a directionally consistent trend for cognitive association of a proxy single-nucleotide polymorphism (SNP), rs3783332. Although the DEGS2 gene is expressed in human brain, it is unknown how DEGS2 expression varies during human life and whether it is affected by psychiatric disorders and genetic variants. To address these questions, we examined DEGS2 messenger RNA using next-generation sequencing in postmortem dorsolateral prefrontal cortical tissue from a total of 418 Caucasian samples including patients with schizophrenia, bipolar disorder and major depressive disorder. DEGS2 is expressed at very low levels prenatally and increases gradually from birth to adolescence and consistently expressed across adulthood. Rs3783332 genotype was significantly associated with the expression across all subjects (F3,348=10.79, P=1.12 × 10(-)(3)), particularly in control subjects (F1,87=13.14, P=4.86 × 10(-4)). Similar results were found with rs715799 genotype. The carriers of the risk-associated minor allele at both loci showed significantly lower expression compared with subjects homozygous for the non-risk major allele and this was a consistent finding across all diagnostic groups. DEGS2 expression showed no association with diagnostic status after correcting for multiple testing (P>0.05). Our findings demonstrate that a SNP showing genome-wide association study significant association with cognition in schizophrenia is also associated with regulation of DEGS2 expression, implicating a molecular mechanism for the clinical association.

Dominance genetic variation contributes little to the missing heritability for human complex traits

For human complex traits, non-additive genetic variation has been invoked to explain "missing heritability," but its discovery is often neglected in genome-wide association studies. Here we propose a method of using SNP data to partition and estimate the proportion of phenotypic variance attributed to additive and dominance genetic variation at all SNPs (hSNP(2) and δSNP(2)) in unrelated individuals based on an orthogonal model where the estimate of hSNP(2) is independent of that of δSNP(2). With this method, we analyzed 79 quantitative traits in 6,715 unrelated European Americans. The estimate of δSNP(2) averaged across all the 79 quantitative traits was 0.03, approximately a fifth of that for additive variation (average hSNP(2) = 0.15). There were a few traits that showed substantial estimates of δSNP(2), none of which were replicated in a larger sample of 11,965 individuals. We further performed genome-wide association analyses of the 79 quantitative traits and detected SNPs with genome-wide significant dominance effects only at the ABO locus for factor VIII and von Willebrand factor. All these results suggest that dominance variation at common SNPs explains only a small fraction of phenotypic variation for human complex traits and contributes little to the missing narrow-sense heritability problem.

The evolution and adaptive potential of transcriptional variation in sticklebacks--signatures of selection and widespread heritability

Evidence implicating differential gene expression as a significant driver of evolutionary novelty continues to accumulate, but our understanding of the underlying sources of variation in expression, both environmental and genetic, is wanting. Heritability in particular may be underestimated when inferred from genetic mapping studies, the predominant "genetical genomics" approach to the study of expression variation. Such uncertainty represents a fundamental limitation to testing for adaptive evolution at the transcriptomic level. By studying the inheritance of expression levels in 10,495 genes (10,527 splice variants) in a threespine stickleback pedigree consisting of 563 individuals, half of which were subjected to a thermal treatment, we show that 74-98% of transcripts exhibit significant additive genetic variance. Dominance variance is also prevalent (41-99% of transcripts), and genetic sources of variation seem to play a more significant role in expression variance in the liver than a key environmental variable, temperature. Among-population comparisons suggest that the majority of differential expression in the liver is likely due to neutral divergence; however, we also show that signatures of directional selection may be more prevalent than those of stabilizing selection. This predominantly aligns with the neutral model of evolution for gene expression but also suggests that natural selection may still act on transcriptional variation in the wild. As genetic variation both within- and among-populations ultimately defines adaptive potential, these results indicate that broad adaptive potential may be found within the transcriptome.

Evolutionary genetic bases of longevity and senescence

Senescence, as a time-dependent developmental process, affects all organisms at every stage in their development and growth. During this process, genetic, epigenetic and environmental factors are known to introduce a wide range of variation for longevity among individuals. As an important life-history trait, longevity shows ontogenetic relationships with other complex traits, and hence may be viewed as a composite trait. Factors that influence the origin and maintenance of diversity of life are ultimately governed by Darwinian processes. Here we review evolutionary genetic mechanisms underlying longevity and senescence in humans from a life-history and genotype-epigenetic-phenotype (G-E-P) map prospective. We suggest that synergistic and cascading effects of cis-ruptive mechanisms in the genome, and epigenetic disruptive processes in relation to environmental factors may lead to sequential slippage in the G-E-P space. These mechanisms accompany age, stage and individual specific senescent processes, influenced by positive pleiotropy of certain genes, superior genome integrity, negative-frequency dependent selection and other factors that universally regulate rarity in nature. Finally we interpret life span as an inherent property of self-organizing systems that, accordingly, maintain species-specific limits for the entire complex of fitness traits. We conclude that Darwinian approaches provide unique opportunities to discover the biological bases of longevity as well as devise individual specific medical or other interventions toward improving health span.