Joint Association Analysis Research Articles

Use of multivariate factor analysis of detailed milk fatty acid profile to perform a genome-wide association study in Italian Simmental and Italian Holstein.

Milk fatty acid (FA) profile is a clear example of complex and multiple correlated traits whose genetic basis is difficult to assess. Although genome-wide association (GWA) studies have been successful in the identification of significant genetic variants for complex traits, when correlated phenotypes are analysed separately, the outcomes are difficult to compare and interpret in a metabolic context. Here, we performed a multivariate factor analysis (MFA) on Italian Simmental and Italian Holstein milk fat profiles to extract latent unobserved factors able to explain correlation structure and common metabolic function among different FAs. Individual factor scores obtained by MFA were used to perform a single-SNP based GWA. In both breeds, MFA was able to extract ten latent factors with specific biological meaning, notably: de novo synthesis, desaturation activity and biohydrogenation. The GWA result confirmed the increased power of joint association analysis on multiple correlated traits and allowed us to identify major candidate genes with well-documented function consistent with the metabolic classification of factors obtained, such as DGAT1, FASN and SCD.

Urine Glucose Excretion Attenuates the Association Between Lipid Accumulation Product and Serum Uric Acid in Subjects with Prediabetes.

PurposeObesity is known to be strongly associated with hyperuricemia. Moreover, the impact of urine glucose excretion (UGE) on serum uric acid (UA) levels has gained much more attention in recent years. Yet concern is raised about whether UGE influences the relationship between obesity and hyperuricemia. The aim of this study was to assess the effect of UGE on the association between lipid accumulation product (LAP), a novel marker of visceral adipose accumulation, and UA in subjects with prediabetes.Materials and MethodsData were obtained from a cross-sectional study. A total of 3645 subjects with prediabetes were included in the present study. The separate and joint associations of LAP and UGE with hyperuricemia were examined using logistic regression analyses.ResultsLAP was positively associated with UA in both genders. Subgroup analysis based on UGE revealed that the association was strongest in subjects with low UGE (r = 0.328, p < 0.001), whereas the positive association was weakened, but still remained significant in subjects with moderate and high UGE. High LAP was significantly associated with an increased odds ratio for hyperuricemia after adjustment for potential confounders in the overall population (OR = 2.07, 95% CI: 1.66–2.58, p < 0.001). However, a downward trend in odds ratios for hyperuricemia was observed across UGE categories. In addition, the joint association analysis confirmed that the relationship between LAP and hyperuricemia was attenuated by UGE.ConclusionThe positive association between LAP and UA appears to be attenuated by UGE, indicating that promoting UGE may be an effective strategy for controlling UA levels, especially for people with obesity who are at increased risk for hyperuricemia.

Brain MRI Phenotypes Associated with Polymorphisms at or near the MAPT (Microtubule Associated Protein Tau) and COASY (Coenzyme A Synthase) Genes

Abstract Objectives Impaired metal homeostasis in the brain has been reported in neurodegenerative diseases such as Alzheimer's Disease and Parkinson's Disease and can be detected using magnetic resonance imaging (MRI). Many factors, including genetics, affect brain metal accumulation and the risk of neurodegeneration or premature brain aging. Our objective is to uncover independent association signals at chromosome 17 associated with susceptibility-weighted MRI (swMRI) intensities in the caudate and putamen. Methods Conditional and joint association analysis by GCTA-COJO was performed on the summary statistics dataset from a genome-wide association study (GWAS) on brain swMRI phenotypes from 7778 healthy adults of European descent aged 40–69 years from the UK BioBank. Independent association signals (single nucleotide polymorphisms or SNPs) found by GCTA-COJO were grouped further by SNPclip to identify additional candidate SNPs in high linkage disequilibrium (LD) (R2 &amp;gt; 0.8, P &amp;lt; 5.0 × 10–8) with the independent lead/index SNP at each signal. Correlations were performed to show the similarity of impacts (beta) of SNPs significantly associated with MRI differences in both caudate and putamen at one locus. Results There are 2 independent and distinct loci at chromosome 17 associated with MRI differences in the putamen, whereas there is only 1 locus for the caudate. One locus for the putamen is associated with the MAPT H2 haplotype, which has been linked to different neurological diseases. Another locus for both putamen and caudate is located near the genes COASY, NAGLU, and MLX, which are all plausible candidate genes for brain metal accumulation. For one locus identified by GCTA-COJO, there are 19 SNPs in LD (R2 &amp;gt; 0.8) and overlapping in both the caudate and the putamen. The effects (beta) of the SNPs in the putamen and caudate are highly correlated across both brain regions (R2 = 0.907, P-value = 0.000921). Conclusions There is likely 1 signal from the COASY/NAGLU locus on chromosome 17 affecting both brain regions. The signal from the MAPT locus only affects the MRI pattern seen in the putamen. These results connect genetics with brain MRI patterns and an individual's risk for neurodegenerative diseases. Funding Sources This work was funded by grants from the Oklahoma Center for the Advancement of Science &amp; Technology and the Oklahoma Agricultural Experiment Station.

Brain MRI Phenotypes Associated with Polymorphisms at or Near the SLC39A8 (ZIP8) Metal Transporter Gene

Abstract Objectives The SLC39A8 gene encodes a divalent metal transporter, ZIP8. ZIP8 polymorphisms are associated with pleiotropic effects including altered risks for schizophrenia. Our objective is to determine the different brain MRI phenotypes associated at or near the SLC39A8 (ZIP8) genetic locus using a phenome-wide association (PheWAS) approach followed by joint and conditional association analysis. Methods Using the summary statistics database containing brain MRI genome-wide association study (GWAS) data, we systematically selected all brain MRI phenotypes which were associated with single-nucleotide polymorphisms (SNPs) within 1 million bp of the SLC39A8 genetic locus, as defined as reaching a P-value significance cutoff of P &amp;lt; 5.0 × 10–8. For all brain MRI phenotypes reaching significance, we used GCTA-COJO to determine the number of independent association signals using settings of P &amp;lt; 1.0 × 10–5 and a window of 10 million base pairs. Using SNPclip and the European 1000 Genomes linkage panel with a linkage disequilibrium cutoff of r2 &amp;gt; 0.8, we identified SNP candidates at each index SNP. Linkage equilibrium for brain phenotypes with multiple independent signals was confirmed by LDpair. Results We identified 25 brain MRI phenotypes that vary due to MRI type and brain region that all contain a SNP associated with the SLC39A8 locus. All of these datasets have at least 1 index SNP directly labeling or in high linkage disequilibrium with rs13107325, which encodes a missense mutation in the SLC39A8 (ZIP8). Among the 25 datasets, an additional 4 association signals were identified by GCTA-COJO and confirmed to be in linkage equilibrium with rs13107325 using LDpair. For these additional association signals, probable causative SNPs were identified from the index SNP using SNPclip. Conclusions From the 25 brain MRI phenotypes, we identified new probable causative SNPs in addition to a previously reported missense SNP (rs13107325) associated with schizophrenia. This study provides leads into how SNPs in genes involved in trace metal transport influence brain structures and affect risks for schizophrenia. Funding Sources This work was funded by grants from the Oklahoma Center for the Advancement of Science and Technology and the Oklahoma Agricultural Experiment Station.

Twelve New Genomic Loci Associated With Bone Mineral Density.

Aiming to identify more genomic loci associated with bone mineral density (BMD), we conducted a joint association analysis of 2 genome-wide association study (GWAS) by the integrative association method multi-trait analysis of GWAS (MTAG). The first one is the single GWAS of estimated heel BMD (eBMD) in the UK biobank (UKB) cohort (N = 426,824), and the second one is the GWAS meta-analysis of total body BMD (TB-BMD) in 66,628 participants from 30 studies. Approximate conditional association analysis was performed in the identified novel loci to identify secondary association signal. Statistical fine-mapping was conducted to prioritize plausible credible risk variants (CRVs). Candidate genes were prioritized based on the analyses of cis- expression quantitative trait locus (cis-eQTL) and cis-protein QTL (cis-pQTL) information as well as the functional category of the SNP. By integrating the information carried in over 490,000 participants, this largest joint analysis of BMD GWAS identified 12 novel genomic loci at the genome-wide significance level (GWS, p = 5.0 × 10−8), nine of which were for eBMD and four were for TB-BMD, explaining an additional 0.11 and 0.23% heritability for the two traits, respectively. These loci include 1p33 (lead SNP rs10493130, peBMD = 3.19 × 10−8), 5q13.2 (rs4703589, peBMD = 4.78 × 10−8), 5q31.3 (rs9324887, pTB−BMD = 1.36 × 10−9), 6p21.32 (rs6905837, peBMD = 3.32 × 10−8), 6q14.1 (rs10806234, peBMD = 2.63 × 10−8), 7q21.11 (rs10806234, pTB−BMD = 3.37 × 10−8), 8q24.12 (rs11995866, peBMD = 6.72 × 10−9), 12p13.31 (rs1639122, peBMD = 4.43 × 10−8), 12p12.1 (rs58489179, peBMD = 4.74 × 10−8), 12q24.23 (rs75499226, peBMD = 1.44 × 10−8), 19q13.31 (rs7255083, pTB−BMD = 2.18 × 10−8) and 22q11.23 (rs13056137, pTB−BMD = 2.54 × 10−8). All lead SNPs in these 12 loci are nominally significant in both original studies as well as consistent in effect direction between them, providing solid evidence of replication. Approximate conditional analysis identified one secondary signal in 5q13.2 (rs11738874, pconditional = 5.06 × 10−9). Statistical fine-mapping analysis prioritized 269 CRVs. A total of 65 candidate genes were prioritized, including those with known biological function to bone development (such as FGF1, COL11A2 and DEPTOR). Our findings provide novel insights into a better understanding of the genetic mechanism underlying bone development as well as candidate genes for future functional investigation.

Urine metals concentrations and dyslexia among children in China

BackgroundExposures to heavy metals has been linked to many developmental disorders. However, few studies have investigated the effects of exposure and co-exposure to metals on dyslexia, especially with regard to dyslexics in China. ObjectivesTo investigate the association between urine metal concentrations and dyslexia in a case-control study among children in China. MethodsA case-control study was conducted in the Tongji Reading Environment and Dyslexia (READ) research program. A total of 228 dyslexics and 228 controls were matched on gender, age (within one year), and grade. The concentrations of 21 urine metals were measured by an inductively coupled plasma-mass spectrometer (ICP-MS). Multivariable conditional logistic regression models were applied to estimate the adjusted odds ratios (ORs) of Chinese dyslexia. ResultsAfter adjusting for potential confounders, two metals (selenium and argentum) were significantly associated with dyslexia in single-metal multivariable models. Argentum was positively associated with the risk of dyslexia, while selenium was negatively associated. In the multiple-metal model, compared with extreme quartiles, the ORs (95% CIs) for selenium was 0.36 (95% CI: 0.14–0.93; Ptrend = 0.029). In the joint association analysis, children with higher levels of urine argentum and lower level of urine selenium had a significantly higher risk of dyslexia than those with low levels of argentum and selenium (OR = 5.06, 95% CI: 1.67–18.84). ConclusionsThe findings suggest that exposure to selenium and argentum may relate to dyslexia in China. Longitudinal studies are needed to further evaluate these relationships and investigate potential mechanisms.

Three Novel Loci for Infant Head Circumference Identified by a Joint Association Analysis.

As an important trait at birth, infant head circumference (HC) is associated with a variety of intelligence- and mental-related conditions. Despite being dominated by genetics, the mechanism underlying the variation of HC is poorly understood. Aiming to uncover the genetic basis of HC, we performed a genome-wide joint association analysis by integrating the genome-wide association summary statistics of HC with that of its two related traits, birth length and birth weight, using a recently developed integrative method, multitrait analysis of genome-wide association (MTAG), and performed in silico replication in an independent sample of intracranial volume (N = 26,577). We then conducted a series of bioinformatic investigations on the identified loci. Combining the evidence from both the MTAG analysis and the in silico replication, we identified three novel loci at the genome-wide significance level (α = 5.0 × 10−8): 3q23 [lead single nucleotide polymorphism (SNP) rs9846396, p MTAG = 3.35 × 10−8, p replication = 0.01], 7p15.3 (rs12534093, p MTAG = 2.00 × 10−8, p replication = 0.004), and 9q33.3 (rs7048271 p MTAG = 9.23 × 10−10, p replication = 1.14 × 10−4). Each of the three lead SNPs was associated with at least one of eight brain-related traits including intelligence and educational attainment. Credible risk variants, defined as those SNPs located within 500 kb of the lead SNP and with p values within two orders of magnitude of the lead SNP, were enriched in DNase I hypersensitive site region in brain. Nine candidate genes were prioritized at the three novel loci using multiple sources of information. Gene set enrichment analysis identified one associated pathway GO:0048009, which participates in the development of nervous system. Our findings provide useful insights into the genetic basis of HC and the relationship between brain growth and mental health.

Plasma metals and cardiovascular disease in patients with type 2 diabetes

BackgroundMetals exposure from natural environment and pollution have been linked to cardiovascular disease (CVD). However, whether associations existing between plasma multiple metals and incident cardiovascular disease in patients with type 2 diabetes (T2D) is unknown. ObjectivesWe conducted a prospective cohort study to investigate whether plasma levels of metals are associated with incident CVD risk in patients with T2D. MethodsIn a prospective study of 3897 type 2 diabetes embedded in the Dongfeng–Tongji cohort, fasting blood samples were collected in 2008 at baseline and in 2013 in the first follow-up period. Plasma concentrations of 23 metals were measured by inductively coupled plasma mass spectrometry (ICP-MS). The associations between plasma metal concentrations and CVD risk in patients with T2D were investigated with Cox proportional hazards models. ResultsDuring an average of 6.2 years follow-up, 1114 participants developed CVD. In the single-metal models adjusting for established cardiovascular risk factors, plasma zinc and selenium levels were negatively and strontium was positively associated with incident CVD risk in patients with T2D. Similar results were obtained in the multiple-metal model, the HRs (95% CIs) for zinc, selenium, and strontium comparing extreme quartiles were 0.78 (95% CI: 0.65–0.93; P trend = 0.011), 0.76 (95% CI: 0.64–0.91; P trend = 0.001), and 1.51 (95% CI: 1.26–1.81; P trend <0.001), respectively. In the joint association analyses of two metals, individuals with high plasma levels of zinc and selenium had significantly lower risk of incident CVD in patients with T2D than those with low levels (HR = 0.77, 95% CI: 0.65–0.91). ConclusionsThe present study suggested that plasma levels of zinc and selenium had an inverse association with incident CVD risk in patients with T2D, while strontium had a positive correlation. Plasma zinc and selenium combinedly decreased incident CVD risk in patients with T2D. Further research is still needed to verify these findings in other populations.

Considering Genetic Heterogeneity in the Association Analysis Finds Genes Associated With Nicotine Dependence.

While substantial progress has been made in finding genetic variants associated with nicotine dependence (ND), a large proportion of the genetic variants remain undiscovered. The current research focuses have shifted toward uncovering rare variants, gene-gene/gene-environment interactions, and structural variations predisposing to ND, the impact of genetic heterogeneity in ND has been nevertheless paid less attention. The study of genetic heterogeneity in ND not only could enhance the power of detecting genetic variants with heterogeneous effects in the population but also improve our understanding of genetic etiology of ND. As an initial step to understand genetic heterogeneity in ND, we applied a newly developed heterogeneity weighted U (HWU) method to 26 ND-related genes, investigating heterogeneous effects of these 26 genes in ND. We found no strong evidence of genetic heterogeneity in genes such as CHRNA5. However, results from our analysis suggest heterogeneous effects of CHRNA6 and CHRNB3 on nicotine dependence in males and females. Following the gene-based analysis, we further conduct a joint association analysis of two gene clusters, CHRNA5-CHRNA3-CHRNB4 and CHRNB3-CHRNA6. While both CHRNA5-CHRNA3-CHRNB4 and CHRNB3-CHRNA6 clusters are significantly associated with ND, there is a much stronger association of CHRNB3-CHRNA6 with ND when considering heterogeneous effects in gender (p-value = 2.11E-07).

Joint Association Analysis Identified 18 New Loci for Bone Mineral Density.

Bone mineral density (BMD) at various skeletal sites have shared genetic determinants. In the present study, aiming to identify shared loci associated with BMD, we conducted a joint association study of a genomewide association study (GWAS) and a meta-analysis of BMD at different skeletal sites: (i) a single GWAS of heel BMD in 142,487 individuals from the UK Biobank, and (ii) a meta-analysis of 30 GWASs of total body (TB) BMD in 66,628 individuals from the Genetic Factors for Osteoporosis (GEFOS) Consortium. The genetic correlation coefficient of the two traits was estimated to be 0.57. We performed joint association analysis with a recently developed statistical method multi-trait analysis of GWAS (MTAG) to account for trait heterogeneity and sample overlap. The joint association analysis combining samples of up to 209,115 individuals identified 18 novel loci associated with BMD at the genomewide significance level (α = 5.0 × 10-8 ), explaining an additional 0.43% and 0.60% of heel-BMD and TB-BMD heritability, respectively. The vast majority of the identified lead SNPs or their proxies exerted local expression quantitative trait loci (cis-eQTL) activity. Credible risk variants, defined as those SNPs located within 500 kilobases (kb) of the lead SNP and with p values within two orders of magnitude of the lead SNP, were enriched in transcription factor binding sites (p = 3.58 × 10-4 ) and coding regions (p = 5.71 × 10-4 ). Fifty-six candidate genes were prioritized at these novel loci using multiple sources of information, including several genes being previously reported to play a role in bone biology but not reported in previous GWASs (PPARG, FBN2, DEF6, TNFRSF19, and NFE2L1). One newly identified gene, SCMH1, was shown to upregulate the expression of several bone biomarkers, including alkaline phosphatase (ALP), collagen type 1 (COL-I), osteocalcin (OCN), osteopontin (OPN), and runt-related transcription factor 2 (RUNX2), in mouse osteoblastic MC3T3-E1 cells, highlighting its regulatory role in bone formation. Our results may provide useful candidate genes for future functional investigations. © 2019 American Society for Bone and Mineral Research.

The eMERGE genotype set of 83,717 subjects imputed to ~40 million variants genome wide and association with the herpes zoster medical record phenotype.

The Electronic Medical Records and Genomics (eMERGE) network is a network of medical centers with electronic medical records linked to existing biorepository samples for genomic discovery and genomic medicine research. The network sought to unify the genetic results from 78 Illumina and Affymetrix genotype array batches from 12 contributing medical centers for joint association analysis of 83,717 human participants. In this report, we describe the imputation of eMERGE results and methods to create the unified imputed merged set of genome‐wide variant genotype data. We imputed the data using the Michigan Imputation Server, which provides a missing single‐nucleotide variant genotype imputation service using the minimac3 imputation algorithm with the Haplotype Reference Consortium genotype reference set. We describe the quality control and filtering steps used in the generation of this data set and suggest generalizable quality thresholds for imputation and phenotype association studies. To test the merged imputed genotype set, we replicated a previously reported chromosome 6 HLA‐B herpes zoster (shingles) association and discovered a novel zoster‐associated loci in an epigenetic binding site near the terminus of chromosome 3 (3p29).

Joint linkage and linkage disequilibrium mapping reveals association of BRMS1L with total teat number in a large intercross between Landrace and Korean native pigs

Previously, we reported a quantitative trait locus (QTL) that affects total teat number (TTN) on pig chromosome 7 (SSC7) in a large F2 intercross population between Landrace and Korean native pigs. The aim of this study was to refine the QTL associated with TTN and to identify positional candidate gene(s) within the refined genomic region. TTN was recorded in 1105 F2 progeny. All experimental animals were genotyped using 980 informative single nucleotide polymorphism (SNP) markers located on SSC7. A haplotype-based linkage and association analysis using the PHASEBOOK programme was applied to refine the QTL region. Additionally, linear mixed-effect models were used to assess the effect of a positional candidate gene on TTN and other economically important traits [i.e., thoracic vertebrae number (THO), carcass body length (CBL) and weight (CW), back fat thickness (BFT) and intramuscular fat content (IMF) in loin muscles]. Joint linkage and association analysis refined the critical region to a 4.50 Mb region that included a novel positional candidate gene, BRMS1L that encodes the breast cancer metastasis-suppressor 1-like protein, which could possibly be implicated in normal mammary gland development. Significant association of an SNP marker (g. −1087 G > A) in the 5′-flanking region of BRMS1L with TTN (P = 1.10 × 10−8), THO (P = 5.80 × 10−4), and CBL (P = 0.038) was observed. Based on these data, we propose BRMS1L as a positional candidate gene for TTN in pigs. After validation of the association in other independent populations, these results could be useful in optimizing breeding programmes that improve TTN and other economically important traits in swine.

Joint analysis of genetic and epigenetic data using a conditional autoregressive model

BackgroundRapidly evolving high-throughput technology has made it cost-effective to collect multilevel omic data in clinical and biological studies. Different types of omic data collected from these studies provide both shared and complementary information, and can be integrated into association analysis to enhance the power of identifying novel disease-associated biomarkers. To model the joint effect of genetic markers and DNA methylation on the phenotype of interest, we propose a joint conditional autoregressive (JCAR) model. A linear score test is used for hypothesis testing and the corresponding p value can be obtained using the Davies method.ResultsThe JCAR model was applied to the GAW20 data from the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) study. In our application of the JCAR model, we consider a baseline model and a full model. In the baseline model, we consider 3 different scenarios: a model with only genetic information, a model with only DNA methylation information at visit 2, and a model using both genetic and DNA methylation information at visit 2. For the full model, we consider both genetic and DNA methylation information at visit 2 and visit 4. The top 10 significant genes are reported for each model. Based on the results, we found that the gene MYO3B was significant as long as the methylation information was considered in the analysis.ConclusionsJCAR is a useful tool for joint association analysis of genetic and epigenetic data. It is easy to implement and is computationally efficient. It can also be extended to analyze other types of omic data.

Two-way mixed-effects methods for joint association analysis using both host and pathogen genomes

Infectious diseases are often affected by specific pairings of hosts and pathogens and therefore by both of their genomes. The integration of a pair of genomes into genome-wide association mapping can provide an exquisitely detailed view of the genetic landscape of complex traits. We present a statistical method, ATOMM (Analysis with a Two-Organism Mixed Model), that maps a trait of interest to a pair of genomes simultaneously; this method makes use of whole-genome sequence data for both host and pathogen organisms. ATOMM uses a two-way mixed-effect model to test for genetic associations and cross-species genetic interactions while accounting for sample structure including interactions between the genetic backgrounds of the two organisms. We demonstrate the applicability of ATOMM to a joint association study of quantitative disease resistance (QDR) in the Arabidopsis thaliana-Xanthomonas arboricola pathosystem. Our method uncovers a clear host-strain specificity in QDR and provides a powerful approach to identify genetic variants on both genomes that contribute to phenotypic variation.

Using genome wide association studies to identify common QTL regions in three different genetic backgrounds based on Iberian pig breed.

One of the major limitation for the application of QTL results in pig breeding and QTN identification has been the limited number of QTL effects validated in different animal material. The aim of the current work was to validate QTL regions through joint and specific genome wide association and haplotype analyses for growth, fatness and premier cut weights in three different genetic backgrounds, backcrosses based on Iberian pigs, which has a major role in the analysis due to its high productive relevance. The results revealed nine common QTL regions, three segregating in all three backcrosses on SSC1, 0–3 Mb, for body weight, on SSC2, 3–9 Mb, for loin bone-in weight, and on SSC7, 3 Mb, for shoulder weight, and six segregating in two of the three backcrosses, on SSC2, SSC4, SSC6 and SSC10 for backfat thickness, shoulder and ham weights. Besides, 18 QTL regions were specifically identified in one of the three backcrosses, five identified only in BC_LD, seven in BC_DU and six in BC_PI. Beyond identifying and validating QTL, candidate genes and gene variants within the most interesting regions have been explored using functional annotation, gene expression data and SNP identification from RNA-Seq data. The results allowed us to propose a promising list of candidate mutations, those identified in PDE10A, DHCR7, MFN2 and CCNY genes located within the common QTL regions and those identified near ssc-mir-103-1 considered PANK3 regulators to be further analysed.

IDENTIFICATION OF DISCRIMINATIVE IMAGING PROTEOMICS ASSOCIATIONS IN ALZHEIMER'S DISEASE VIA A NOVEL SPARSE CORRELATION MODEL.

Brain imaging and protein expression, from both cerebrospinal fluid and blood plasma, have been found to provide complementary information in predicting the clinical outcomes of Alzheimer's disease (AD). But the underlying associations that contribute to such a complementary relationship have not been previously studied yet. In this work, we will perform an imaging proteomics association analysis to explore how they are related with each other. While traditional association models, such as Sparse Canonical Correlation Analysis (SCCA), can not guarantee the selection of only disease-relevant biomarkers and associations, we propose a novel discriminative SCCA (denoted as DSCCA) model with new penalty terms to account for the disease status information. Given brain imaging, proteomic and diagnostic data, the proposed model can perform a joint association and multi-class discrimination analysis, such that we can not only identify disease-relevant multimodal biomarkers, but also reveal strong associations between them. Based on a real imaging proteomic data set, the empirical results show that DSCCA and traditional SCCA have comparable association performances. But in a further classification analysis, canonical variables of imaging and proteomic data obtained in DSCCA demonstrate much more discrimination power toward multiple pairs of diagnosis groups than those obtained in SCCA.

Joint association analysis of a binary and a quantitative trait in family samples.

In recent years, improved genotyping and sequencing technologies have enabled the discovery of new loci associated with various diseases or traits. For instance, by testing the association with each single-nucleotide variant (SNV) separately, genome-wide association studies (GWAS) have achieved tremendous success in identifying SNVs associated with specific traits. However, little is known about the common genetic basis of multiple traits owing to lack of efficient methods. With the use of extended quasi-likelihood, a Wald test has been proposed to perform a bivariate analysis of a continuous and a binary trait in unrelated samples. However, owing to its low computational efficiency, it has not been implemented in real applications to large-scale genetic studies. In this paper, we propose an efficient bivariate robust score test for two traits, one continuous and one binary, based on extended generalized estimating equations. Our approach is applicable to both family-based and unrelated study designs and can be extended to test the association of multiple traits. Our simulation studies demonstrate the type-I error rate of our approach is well controlled in all minor allele frequency (MAF) scenarios, with MAF ranging from 1 to 30%, and the method is more powerful in certain MAF scenarios than univariate testing with correction for multiple testing. Because of the computational advantage of score tests, our approach is readily applicable to GWAS or sequencing studies. Finally, we present a real application to uncover genetic variants associated with body mass index and type-2 diabetes in the Framingham Heart Study.

Genome-wide joint analysis of single-nucleotide variant sets and gene expression for hypertension and related phenotypes.

BackgroundWith the advance of next-generation sequencing technologies, the study of rare variants in targeted genome regions or even the whole genome becomes feasible. Nevertheless, the massive amount of sequencing data brings great computational and statistical challenges for association analyses. Aside from sequencing variants, other high-throughput omic data (eg, gene expression data) also become available, and can be incorporated into association analysis for better modeling and power improvement. This motivates the need of developing computationally efficient and powerful approaches to model the joint associations of multilevel omic data with complex human diseases.MethodsA similarity-based weighted U approach is used to model the joint effect of sequencing variants and gene expression. Using a Mexican American sample provided by Genetic Analysis Workshop 19 (GAW19), we performed a whole-genome joint association analysis of sequencing variants and gene expression with systolic (SBP) and diastolic blood pressure (DBP) and hypertension (HTN) phenotypes.ResultsThe whole-genome joint association analysis was completed in 80 min on a high-performance personal computer with an i7 4700 CPU and 8 GB memory. Although no gene reached statistical significance after adjusting for multiple testing, some top-ranked genes attained a high significance level and may have biological plausibility to hypertension-related phenotypes.ConclusionsThe weighted U approach is computationally efficient for high-dimensional data analysis, and is capable of integrating multiple levels of omic data into association analysis. Through a real data application, we demonstrate the potential benefit of using the new approach for joint association analysis of sequencing variants and gene expression.

0311 Joint association analysis of additive and non-additive genomic effects for growth and carcass traits of beef cattle

0311 Joint association analysis of additive and non-additive genomic effects for growth and carcass traits of beef cattle

Detecting Gene-Gene Interactions Associated with Multiple Complex Traits with U-Statistics.

Many complex diseases, such as psychiatric and behavioral disorders, are commonly characterized through various measurements that reflect physical, behavioral and psychological aspects of diseases. While it remains a great challenge to find a unified measurement to characterize a disease, the available multiple phenotypes can be analyzed jointly in the genetic association study. Simultaneously testing these phenotypes has many advantages, including considering different aspects of the disease in the analysis, and utilizing correlated phenotypes to improve the power of detecting disease-associated variants. Furthermore, complex diseases are likely caused by the interplay of multiple genetic variants through complicated mechanisms. Considering gene-gene interactions in the joint association analysis of complex diseases could further increase our ability to discover genetic variants involving complex disease pathways. In this article, we propose a stepwise U-test for joint association analysis of multiple loci and multiple phenotypes. Through simulations, we demonstrated that testing multiple phenotypes simultaneously could attain higher power than testing one single phenotype at a time, especially when there are shared genes contributing to multiple phenotypes. We also illustrated the proposed method with an application to Nicotine Dependence (ND), using datasets from the Study of Addition, Genetics and Environment (SAGE). The joint analysis of three ND phenotypes identified two SNPs, rs10508649 and rs2491397, and reached a nominal P-value of 3.79e-13. The association was further replicated in two independent datasets with P-values of 2.37e-05 and 7.46e-05.