MRLocus: Identifying causal genes mediating a trait through Bayesian estimation of allelic heterogeneity

Expression quantitative trait loci (eQTL) studies are used to understand the regulatory function of non-coding genome-wide association study (GWAS) risk loci, but colocalization alone does not demonstrate a causal relationship of gene expression affecting a trait. Evidence for mediation, that perturbation of gene expression in a given tissue or developmental context will induce a change in the downstream GWAS trait, can be provided by two-sample Mendelian Randomization (MR). Here, we introduce a new statistical method, MRLocus, for Bayesian estimation of the gene-to-trait effect from eQTL and GWAS summary data for loci with evidence of allelic heterogeneity, that is, containing multiple causal variants. MRLocus makes use of a colocalization step applied to each nearly-LD-independent eQTL, followed by an MR analysis step across eQTLs. Additionally, our method involves estimation of the extent of allelic heterogeneity through a dispersion parameter, indicating variable mediation effects from each individual eQTL on the downstream trait. Our method is evaluated against other state-of-the-art methods for estimation of the gene-to-trait mediation effect, using an existing simulation framework. In simulation, MRLocus often has the highest accuracy among competing methods, and in each case provides more accurate estimation of uncertainty as assessed through interval coverage. MRLocus is then applied to five candidate causal genes for mediation of particular GWAS traits, where gene-to-trait effects are concordant with those previously reported. We find that MRLocus’s estimation of the causal effect across eQTLs within a locus provides useful information for determining how perturbation of gene expression or individual regulatory elements will affect downstream traits. The MRLocus method is implemented as an R package available at https://mikelove.github.io/mrlocus.     

Lung eQTLs to Help Reveal the Molecular Underpinnings of Asthma

Genome-wide association studies (GWAS) have identified loci reproducibly associated with pulmonary diseases; however, the molecular mechanism underlying these associations are largely unknown. The objectives of this study were to discover genetic variants affecting gene expression in human lung tissue, to refine susceptibility loci for asthma identified in GWAS studies, and to use the genetics of gene expression and network analyses to find key molecular drivers of asthma. We performed a genome-wide search for expression quantitative trait loci (eQTL) in 1,111 human lung samples. The lung eQTL dataset was then used to inform asthma genetic studies reported in the literature. The top ranked lung eQTLs were integrated with the GWAS on asthma reported by the GABRIEL consortium to generate a Bayesian gene expression network for discovery of novel molecular pathways underpinning asthma. We detected 17,178 cis- and 593 trans- lung eQTLs, which can be used to explore the functional consequences of loci associated with lung diseases and traits. Some strong eQTLs are also asthma susceptibility loci. For example, rs3859192 on chr17q21 is robustly associated with the mRNA levels of GSDMA (P = 3.55×10⁻¹⁵¹). The genetic-gene expression network identified the SOCS3 pathway as one of the key drivers of asthma. The eQTLs and gene networks identified in this study are powerful tools for elucidating the causal mechanisms underlying pulmonary disease. This data resource offers much-needed support to pinpoint the causal genes and characterize the molecular function of gene variants associated with lung diseases.     

Alzheimer’s Disease Risk Polymorphisms Regulate Gene Expression in the ZCWPW1 and the CELF1 Loci

Late onset Alzheimer’s disease (LOAD) is a genetically complex and clinically heterogeneous disease. Recent large-scale genome wide association studies (GWAS) have identified more than twenty loci that modify risk for AD. Despite the identification of these loci, little progress has been made in identifying the functional variants that explain the association with AD risk. Thus, we sought to determine whether the novel LOAD GWAS single nucleotide polymorphisms (SNPs) alter expression of LOAD GWAS genes and whether expression of these genes is altered in AD brains. The majority of LOAD GWAS SNPs occur in gene dense regions under large linkage disequilibrium (LD) blocks, making it unclear which gene(s) are modified by the SNP. Thus, we tested for brain expression quantitative trait loci (eQTLs) between LOAD GWAS SNPs and SNPs in high LD with the LOAD GWAS SNPs in all of the genes within the GWAS loci. We found a significant eQTL between rs1476679 and PILRB and GATS, which occurs within the ZCWPW1 locus. PILRB and GATS expression levels, within the ZCWPW1 locus, were also associated with AD status. Rs7120548 was associated with MTCH2 expression, which occurs within the CELF1 locus. Additionally, expression of several genes within the CELF1 locus, including MTCH2, were highly correlated with one another and were associated with AD status. We further demonstrate that PILRB, as well as other genes within the GWAS loci, are most highly expressed in microglia. These findings together with the function of PILRB as a DAP12 receptor supports the critical role of microglia and neuroinflammation in AD risk.     

A genome-wide association study of seed protein and oil content in soybean

Background

Association analysis is an alternative to conventional family-based methods to detect the location of gene(s) or quantitative trait loci (QTL) and provides relatively high resolution in terms of defining the genome position of a gene or QTL. Seed protein and oil concentration are quantitative traits which are determined by the interaction among many genes with small to moderate genetic effects and their interaction with the environment. In this study, a genome-wide association study (GWAS) was performed to identify quantitative trait loci (QTL) controlling seed protein and oil concentration in 298 soybean germplasm accessions exhibiting a wide range of seed protein and oil content.

Results

A total of 55,159 single nucleotide polymorphisms (SNPs) were genotyped using various methods including Illumina Infinium and GoldenGate assays and 31,954 markers with minor allele frequency >0.10 were used to estimate linkage disequilibrium (LD) in heterochromatic and euchromatic regions. In euchromatic regions, the mean LD (r ² ) rapidly declined to 0.2 within 360 Kbp, whereas the mean LD declined to 0.2 at 9,600 Kbp in heterochromatic regions. The GWAS results identified 40 SNPs in 17 different genomic regions significantly associated with seed protein. Of these, the five SNPs with the highest associations and seven adjacent SNPs were located in the 27.6-30.0 Mbp region of Gm20. A major seed protein QTL has been previously mapped to the same location and potential candidate genes have recently been identified in this region. The GWAS results also detected 25 SNPs in 13 different genomic regions associated with seed oil. Of these markers, seven SNPs had a significant association with both protein and oil.

Conclusions

This research indicated that GWAS not only identified most of the previously reported QTL controlling seed protein and oil, but also resulted in narrower genomic regions than the regions reported as containing these QTL. The narrower GWAS-defined genome regions will allow more precise marker-assisted allele selection and will expedite positional cloning of the causal gene(s).     

Genetic and Epigenetic Regulation of TOX3 Expression in Breast Cancer

Genome wide association studies (GWAS) have identified low penetrance and high frequency single nucleotide polymorphisms (SNPs) that contribute to genetic susceptibility of breast cancer. The SNPs at 16q12, close to the TOX3 and CASC16 genes, represent one of the susceptibility loci identified by GWAS, showing strong evidence for breast cancer association across various populations. To examine molecular mechanisms of TOX3 regulation in breast cancer, we investigated both genetic and epigenetic factors using cell lines and datasets derived from primary breast tumors available through The Cancer Genome Atlas (TCGA). TOX3 expression is highly up-regulated in luminal subtype tumors compared to normal breast tissues or basal-like tumors. Expression quantitative trait loci (eQTL) analyses revealed significant associations of rs3803662 and rs4784227 genotypes with TOX3 expression in breast tumors. Bisulfite sequencing of four CpG islands in the TOX3 promoter showed a clear difference between luminal and basal-like cancer cell lines. 5-Aza-2’-deoxycytidine treatment of a basal-like cancer cell line increased expression of TOX3. TCGA dataset verified significantly lower levels of methylation of the promoter in luminal breast tumors with an inverse correlation between methylation and expression of TOX3. Methylation QTL (mQTL) analyses showed a weak or no correlation of rs3803662 or rs4784227 with TOX3 promoter methylation in breast tumors, indicating an independent relationship between the genetic and epigenetic events. These data suggest a complex system of TOX3 regulation in breast tumors, driven by germline variants and somatic epigenetic modifications in a subtype specific manner.     

Genetics of sputum gene expression in chronic obstructive pulmonary disease

Previous expression quantitative trait loci (eQTL) studies have performed genetic association studies for gene expression, but most of these studies examined lymphoblastoid cell lines from non-diseased individuals. We examined the genetics of gene expression in a relevant disease tissue from chronic obstructive pulmonary disease (COPD) patients to identify functional effects of known susceptibility genes and to find novel disease genes. By combining gene expression profiling on induced sputum samples from 131 COPD cases from the ECLIPSE Study with genomewide single nucleotide polymorphism (SNP) data, we found 4315 significant cis-eQTL SNP-probe set associations (3309 unique SNPs). The 3309 SNPs were tested for association with COPD in a genomewide association study (GWAS) dataset, which included 2940 COPD cases and 1380 controls. Adjusting for 3309 tests (p<1.5e-5), the two SNPs which were significantly associated with COPD were located in two separate genes in a known COPD locus on chromosome 15: CHRNA5 and IREB2. Detailed analysis of chromosome 15 demonstrated additional eQTLs for IREB2 mapping to that gene. eQTL SNPs for CHRNA5 mapped to multiple linkage disequilibrium (LD) bins. The eQTLs for IREB2 and CHRNA5 were not in LD. Seventy-four additional eQTL SNPs were associated with COPD at p<0.01. These were genotyped in two COPD populations, finding replicated associations with a SNP in PSORS1C1, in the HLA-C region on chromosome 6. Integrative analysis of GWAS and gene expression data from relevant tissue from diseased subjects has located potential functional variants in two known COPD genes and has identified a novel COPD susceptibility locus.     

Refining Susceptibility Loci of Chronic Obstructive Pulmonary Disease with Lung eqtls

Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of mortality worldwide. Recent genome-wide association studies (GWAS) have identified robust susceptibility loci associated with COPD. However, the mechanisms mediating the risk conferred by these loci remain to be found. The goal of this study was to identify causal genes/variants within susceptibility loci associated with COPD. In the discovery cohort, genome-wide gene expression profiles of 500 non-tumor lung specimens were obtained from patients undergoing lung surgery. Blood-DNA from the same patients were genotyped for 1,2 million SNPs. Following genotyping and gene expression quality control filters, 409 samples were analyzed. Lung expression quantitative trait loci (eQTLs) were identified and overlaid onto three COPD susceptibility loci derived from GWAS; 4q31 (HHIP), 4q22 (FAM13A), and 19q13 (RAB4B, EGLN2, MIA, CYP2A6). Significant eQTLs were replicated in two independent datasets (n = 363 and 339). SNPs previously associated with COPD and lung function on 4q31 (rs1828591, rs13118928) were associated with the mRNA expression of HHIP. An association between mRNA expression level of FAM13A and SNP rs2045517 was detected at 4q22, but did not reach statistical significance. At 19q13, significant eQTLs were detected with EGLN2. In summary, this study supports HHIP, FAM13A, and EGLN2 as the most likely causal COPD genes on 4q31, 4q22, and 19q13, respectively. Strong lung eQTL SNPs identified in this study will need to be tested for association with COPD in case-control studies. Further functional studies will also be needed to understand the role of genes regulated by disease-related variants in COPD.     

Limits on the reproducibility of marker associations with southern leaf blight resistance in the maize nested association mapping population

Background

A previous study reported a comprehensive quantitative trait locus (QTL) and genome wide association study (GWAS) of southern leaf blight (SLB) resistance in the maize Nested Association Mapping (NAM) panel. Since that time, the genomic resources available for such analyses have improved substantially. An updated NAM genetic linkage map has a nearly six-fold greater marker density than the previous map and the combined SNPs and read-depth variants (RDVs) from maize HapMaps 1 and 2 provided 28.5 M genomic variants for association analysis, 17 fold more than HapMap 1. In addition, phenotypic values of the NAM RILs were re-estimated to account for environment-specific flowering time covariates and a small proportion of lines were dropped due to genotypic data quality problems. Comparisons of original and updated QTL and GWAS results confound the effects of linkage map density, GWAS marker density, population sample size, and phenotype estimates. Therefore, we evaluated the effects of changing each of these parameters individually and in combination to determine their relative impact on marker-trait associations in original and updated analyses.

Results

Of the four parameters varied, map density caused the largest changes in QTL and GWAS results. The updated QTL model had better cross-validation prediction accuracy than the previous model. Whereas joint linkage QTL positions were relatively stable to input changes, the residual values derived from those QTL models (used as inputs to GWAS) were more sensitive, resulting in substantial differences between GWAS results. The updated NAM GWAS identified several candidate genes consistent with previous QTL fine-mapping results.

Conclusions

The highly polygenic nature of resistance to SLB complicates the identification of causal genes. Joint linkage QTL are relatively stable to perturbations of data inputs, but their resolution is generally on the order of tens or more Mbp. GWAS associations have higher resolution, but lower power due to stringent thresholds designed to minimize false positive associations, resulting in variability of detection across studies. The updated higher density linkage map improves QTL estimation and, along with a much denser SNP HapMap, greatly increases the likelihood of detecting SNPs in linkage with causal variants. We recommend use of the updated genetic resources and results but emphasize the limited repeatability of small-effect associations.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-1068) contains supplementary material, which is available to authorized users.     

Association mapping of morphological traits in wild and captive zebra finches: reliable within,but not between populations

Identifying causal genetic variants underlying heritable phenotypic variation is a long‐standing goal in evolutionary genetics. We previously identified several quantitative trait loci (QTL) for five morphological traits in a captive population of zebra finches (Taeniopygia guttata) by whole‐genome linkage mapping. We here follow up on these studies with the aim to narrow down on the quantitative trait variants (QTN) in one wild and three captive populations. First, we performed an association study using 672 single nucleotide polymorphisms (SNPs) within candidate genes located in the previously identified QTL regions in a sample of 939 wild‐caught zebra finches. Then, we validated the most promising SNP–phenotype associations (n = 25 SNPs) in 5228 birds from four populations. Genotype–phenotype associations were generally weak in the wild population, where linkage disequilibrium (LD) spans only short genomic distances. In contrast, in captive populations, where LD blocks are large, apparent SNP effects on morphological traits (i.e. associations) were highly repeatable with independent data from the same population. Most of those SNPs also showed significant associations with the same trait in other captive populations, but the direction and magnitude of these effects varied among populations. This suggests that the tested SNPs are not the causal QTN but rather physically linked to them, and that LD between SNPs and causal variants differs between populations due to founder effects. While the identification of QTN remains challenging in nonmodel organisms, we illustrate that it is indeed possible to confirm the location and magnitude of QTL in a population with stable linkage between markers and causal variants.     

Meta-eQTL: a tool set for flexible eQTL meta-analysis

Background

Increasing number of eQTL (Expression Quantitative Trait Loci) datasets facilitate genetics and systems biology research. Meta-analysis tools are in need to jointly analyze datasets of same or similar issue types to improve statistical power especially in trans-eQTL mapping. Meta-analysis framework is also necessary for ChrX eQTL discovery.

Results

We developed a novel tool, meta-eqtl, for fast eQTL meta-analysis of arbitrary sample size and arbitrary number of datasets. Further, this tool accommodates versatile modeling, eg. non-parametric model and mixed effect models. In addition, meta-eqtl readily handles calculation of chrX eQTLs.

Conclusions

We demonstrated and validated meta-eqtl as fast and comprehensive tool to meta-analyze multiple datasets and ChrX eQTL discovery. Meta-eqtl is a set of command line utilities written in R, with some computationally intensive parts written in C. The software runs on Linux platforms and is designed to intelligently adapt to high performance computing (HPC) cluster. We applied the novel tool to liver and adipose tissue data, and revealed eSNPs underlying diabetes GWAS loci.     

Candidate Causal Regulatory Effects by Integration of Expression QTLs with Complex Trait Genetic Associations

The recent success of genome-wide association studies (GWAS) is now followed by the challenge to determine how the reported susceptibility variants mediate complex traits and diseases. Expression quantitative trait loci (eQTLs) have been implicated in disease associations through overlaps between eQTLs and GWAS signals. However, the abundance of eQTLs and the strong correlation structure (LD) in the genome make it likely that some of these overlaps are coincidental and not driven by the same functional variants. In the present study, we propose an empirical methodology, which we call Regulatory Trait Concordance (RTC) that accounts for local LD structure and integrates eQTLs and GWAS results in order to reveal the subset of association signals that are due to cis eQTLs. We simulate genomic regions of various LD patterns with both a single or two causal variants and show that our score outperforms SNP correlation metrics, be they statistical (r²) or historical (D''). Following the observation of a significant abundance of regulatory signals among currently published GWAS loci, we apply our method with the goal to prioritize relevant genes for each of the respective complex traits. We detect several potential disease-causing regulatory effects, with a strong enrichment for immunity-related conditions, consistent with the nature of the cell line tested (LCLs). Furthermore, we present an extension of the method in trans, where interrogating the whole genome for downstream effects of the disease variant can be informative regarding its unknown primary biological effect. We conclude that integrating cellular phenotype associations with organismal complex traits will facilitate the biological interpretation of the genetic effects on these traits.     

Protein Interaction Networks Reveal Novel Autism Risk Genes within GWAS Statistical Noise

Genome-wide association studies (GWAS) for Autism Spectrum Disorder (ASD) thus far met limited success in the identification of common risk variants, consistent with the notion that variants with small individual effects cannot be detected individually in single SNP analysis. To further capture disease risk gene information from ASD association studies, we applied a network-based strategy to the Autism Genome Project (AGP) and the Autism Genetics Resource Exchange GWAS datasets, combining family-based association data with Human Protein-Protein interaction (PPI) data. Our analysis showed that autism-associated proteins at higher than conventional levels of significance (P<0.1) directly interact more than random expectation and are involved in a limited number of interconnected biological processes, indicating that they are functionally related. The functionally coherent networks generated by this approach contain ASD-relevant disease biology, as demonstrated by an improved positive predictive value and sensitivity in retrieving known ASD candidate genes relative to the top associated genes from either GWAS, as well as a higher gene overlap between the two ASD datasets. Analysis of the intersection between the networks obtained from the two ASD GWAS and six unrelated disease datasets identified fourteen genes exclusively present in the ASD networks. These are mostly novel genes involved in abnormal nervous system phenotypes in animal models, and in fundamental biological processes previously implicated in ASD, such as axon guidance, cell adhesion or cytoskeleton organization. Overall, our results highlighted novel susceptibility genes previously hidden within GWAS statistical “noise” that warrant further analysis for causal variants.     

Targeted resequencing of GWAS loci reveals novel genetic variants for milk production traits

Background

Genome wide association study (GWAS) has been proven to be a powerful tool for detecting genomic variants associated with complex traits. However, the specific genes and causal variants underlying these traits remain unclear.

Results

Here, we used target-enrichment strategy coupled with next generation sequencing technique to study target regions which were found to be associated with milk production traits in dairy cattle in our previous GWAS. Among the large amount of novel variants detected by targeted resequencing, we selected 200 SNPs for further association study in a population consisting of 2634 cows. Sixty six SNPs distributed in 53 genes were identified to be associated significantly with on milk production traits. Of the 53 genes, 26 were consistent with our previous GWAS results. We further chose 20 significant genes to analyze their mRNA expression in different tissues of lactating cows, of which 15 were specificly highly expressed in mammary gland.

Conclusions

Our study illustrates the potential for identifying causal mutations for milk production traits using target-enrichment resequencing and extends the results of GWAS by discovering new and potentially functional mutations.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-1105) contains supplementary material, which is available to authorized users.     

Prioritizing genetic variants for causality on the basis of preferential linkage disequilibrium

To date, the widely used genome-wide association studies (GWASs) of the human genome have reported thousands of variants that are significantly associated with various human traits. However, in the vast majority of these cases, the causal variants responsible for the observed associations remain unknown. In order to facilitate the identification of causal variants, we designed a simple computational method called the "preferential linkage disequilibrium (LD)" approach, which follows the variants discovered by GWASs to pinpoint the causal variants, even if they are rare compared with the discovery variants. The approach is based on the hypothesis that the GWAS-discovered variant is better at tagging the causal variants than are most other variants evaluated in the original GWAS. Applying the preferential LD approach to the GWAS signals of five human traits for which the causal variants are already known, we successfully placed the known causal variants among the top ten candidates in the majority of these cases. Application of this method to additional GWASs, including those of hepatitis C virus treatment response, plasma levels of clotting factors, and late-onset Alzheimer disease, has led to the identification of a number of promising candidate causal variants. This method represents a useful tool for delineating causal variants by bringing together GWAS signals and the rapidly accumulating variant data from next-generation sequencing.     

mrMLM v4.0.2:An R Platform for Multi-locus Genome-wide Association Studies

Previous studies have reported that some important loci are missed in single-locus genome-wide association studies (GWAS), especially because of the large phenotypic error in field experiments. To solve this issue, multi-locus GWAS methods have been recommended. However, only a few software packages for multi-locus GWAS are available. Therefore, we developed an R software named mrMLM v4.0.2. This software integrates mrMLM, FASTmrMLM, FASTmrEMMA, pLARmEB, pKWmEB, and ISIS EM-BLASSO methods developed by our lab. There are four components in mrMLM v4.0.2, including dataset input, parameter setting, software running, and result output. The fread function in data.table is used to quickly read datasets, especially big datasets, and the doParallel package is used to conduct parallel computation using multiple CPUs. In addition, the graphical user interface software mrMLM.GUI v4.0.2, built upon Shiny, is also available. To confirm the correctness of the aforementioned programs, all the methods in mrMLM v4.0.2 and three widely-used methods were used to analyze real and simulated datasets. The results confirm the superior performance of mrMLM v4.0.2 to other methods currently available. False positive rates are effectively controlled, albeit with a less stringent significance threshold. mrMLM v4.0.2 is publicly available at BioCode (https://bigd.big.ac.cn/biocode/tools/BT007077) or R (https://cran.r-project.org/web/packages/mrMLM.GUI/index.html) as an open-source software.