Enhancing selection of alcohol consumption-associated genes by random forest.

Machine learning methods have been used in identifying omics markers for a variety of phenotypes. We aimed to examine whether a supervised machine learning algorithm can improve identification of alcohol-associated transcriptomic markers. In this study, we analysed array-based, whole-blood derived expression data for 17 873 gene transcripts in 5508 Framingham Heart Study participants. By using the Boruta algorithm, a supervised random forest (RF)-based feature selection method, we selected twenty-five alcohol-associated transcripts. In a testing set (30 % of entire study participants), AUC (area under the receiver operating characteristics curve) of these twenty-five transcripts were 0·73, 0·69 and 0·66 for non-drinkers v. moderate drinkers, non-drinkers v. heavy drinkers and moderate drinkers v. heavy drinkers, respectively. The AUC of the selected transcripts by the Boruta method were comparable to those identified using conventional linear regression models, for example, AUC of 1958 transcripts identified by conventional linear regression models (false discovery rate < 0·2) were 0·74, 0·66 and 0·65, respectively. With Bonferroni correction for the twenty-five Boruta method-selected transcripts and three CVD risk factors (i.e. at P < 6·7e-4), we observed thirteen transcripts were associated with obesity, three transcripts with type 2 diabetes and one transcript with hypertension. For example, we observed that alcohol consumption was inversely associated with the expression of DOCK4, IL4R, and SORT1, and DOCK4 and SORT1 were positively associated with obesity, and IL4R was inversely associated with hypertension. In conclusion, using a supervised machine learning method, the RF-based Boruta algorithm, we identified novel alcohol-associated gene transcripts.

Methylation patterns associated with C-reactive protein in racially and ethnically diverse populations.

Systemic low-grade inflammation is a feature of chronic disease. C-reactive protein (CRP) is a common biomarker of inflammation and used as an indicator of disease risk; however, the role of inflammation in disease is not completely understood. Methylation is an epigenetic modification in the DNA which plays a pivotal role in gene expression. In this study we evaluated differential DNA methylation patterns associated with blood CRP level to elucidate biological pathways and genetic regulatory mechanisms to improve the understanding of chronic inflammation. The racially and ethnically diverse participants in this study were included as 50% White, 41% Black or African American, 7% Hispanic or Latino/a, and 2% Native Hawaiian, Asian American, American Indian, or Alaska Native (total n = 13,433) individuals. We replicated 113 CpG sites from 87 unique loci, of which five were novel (CADM3, NALCN, NLRC5, ZNF792, and cg03282312), across a discovery set of 1,150 CpG sites associated with CRP level (p < 1.2E-7). The downstream pathways affected by DNA methylation included the identification of IFI16 and IRF7 CpG-gene transcript pairs which contributed to the innate immune response gene enrichment pathway along with NLRC5, NOD2, and AIM2. Gene enrichment analysis also identified the nuclear factor-kappaB transcription pathway. Using two-sample Mendelian randomization (MR) we inferred methylation at three CpG sites as causal for CRP levels using both White and Black or African American MR instrument variables. Overall, we identified novel CpG sites and gene transcripts that could be valuable in understanding the specific cellular processes and pathogenic mechanisms involved in inflammation.

Circulating metabolites may illustrate relationship of alcohol consumption with cardiovascular disease.

BACKGROUND: Metabolite signatures of long-term alcohol consumption are lacking. To better understand the molecular basis linking alcohol drinking and cardiovascular disease (CVD), we investigated circulating metabolites associated with long-term alcohol consumption and examined whether these metabolites were associated with incident CVD. METHODS: Cumulative average alcohol consumption (g/day) was derived from the total consumption of beer, wine, and liquor on average of 19 years in 2428 Framingham Heart Study Offspring participants (mean age 56 years, 52% women). We used linear mixed models to investigate the associations of alcohol consumption with 211 log-transformed plasma metabolites, adjusting for age, sex, batch, smoking, diet, physical activity, BMI, and familial relationship. Cox models were used to test the association of alcohol-related metabolite scores with fatal and nonfatal incident CVD (myocardial infarction, coronary heart disease, stroke, and heart failure). RESULTS: We identified 60 metabolites associated with cumulative average alcohol consumption (p < 0.05/211 ≈ 0.00024). For example, 1 g/day increase of alcohol consumption was associated with higher levels of cholesteryl esters (e.g., CE 16:1, beta = 0.023 ± 0.002, p = 6.3e - 45) and phosphatidylcholine (e.g., PC 32:1, beta = 0.021 ± 0.002, p = 3.1e - 38). Survival analysis identified that 10 alcohol-associated metabolites were also associated with a differential CVD risk after adjusting for age, sex, and batch. Further, we built two alcohol consumption weighted metabolite scores using these 10 metabolites and showed that, with adjustment age, sex, batch, and common CVD risk factors, the two scores had comparable but opposite associations with incident CVD, hazard ratio 1.11 (95% CI = [1.02, 1.21], p = 0.02) vs 0.88 (95% CI = [0.78, 0.98], p = 0.02). CONCLUSIONS: We identified 60 long-term alcohol consumption-associated metabolites. The association analysis with incident CVD suggests a complex metabolic basis between alcohol consumption and CVD.

Multi-tissue epigenetic analysis identifies distinct associations underlying insulin resistance and Alzheimer's disease at CPT1A locus.

BACKGROUND: Insulin resistance (IR) is a major risk factor for Alzheimer's disease (AD) dementia. The mechanisms by which IR predisposes to AD are not well-understood. Epigenetic studies may help identify molecular signatures of IR associated with AD, thus improving our understanding of the biological and regulatory mechanisms linking IR and AD. METHODS: We conducted an epigenome-wide association study of IR, quantified using the homeostatic model assessment of IR (HOMA-IR) and adjusted for body mass index, in 3,167 participants from the Framingham Heart Study (FHS) without type 2 diabetes at the time of blood draw used for methylation measurement. We identified DNA methylation markers associated with IR at the genome-wide level accounting for multiple testing (P < 1.1 × 10-7) and evaluated their association with neurological traits in participants from the FHS (N = 3040) and the Religious Orders Study/Memory and Aging Project (ROSMAP, N = 707). DNA methylation profiles were measured in blood (FHS) or dorsolateral prefrontal cortex (ROSMAP) using the Illumina HumanMethylation450 BeadChip. Linear regressions (ROSMAP) or mixed-effects models accounting for familial relatedness (FHS) adjusted for age, sex, cohort, self-reported race, batch, and cell type proportions were used to assess associations between DNA methylation and neurological traits accounting for multiple testing. RESULTS: We confirmed the strong association of blood DNA methylation with IR at three loci (cg17901584-DHCR24, cg17058475-CPT1A, cg00574958-CPT1A, and cg06500161-ABCG1). In FHS, higher levels of blood DNA methylation at cg00574958 and cg17058475 were both associated with lower IR (P = 2.4 × 10-11 and P = 9.0 × 10-8), larger total brain volumes (P = 0.03 and P = 9.7 × 10-4), and smaller log lateral ventricular volumes (P = 0.07 and P = 0.03). In ROSMAP, higher levels of brain DNA methylation at the same two CPT1A markers were associated with greater risk of cognitive impairment (P = 0.005 and P = 0.02) and higher AD-related indices (CERAD score: P = 5 × 10-4 and 0.001; Braak stage: P = 0.004 and P = 0.01). CONCLUSIONS: Our results suggest potentially distinct epigenetic regulatory mechanisms between peripheral blood and dorsolateral prefrontal cortex tissues underlying IR and AD at CPT1A locus.

Alcohol consumption and epigenetic age acceleration across human adulthood.

The alcohol-associated biological aging remains to be studied across adulthood. We conducted linear regression analyses to investigate the associations between alcohol consumption and two DNA methylation-based biological age acceleration metrics in 3823 Framingham Heart Study participants (24-92 years and 53.8% women) adjusting for covariates. We also investigated whether the two epigenetic aging metrics mediated the association of alcohol consumption with hypertension. We found that higher long-term average alcohol consumption was significantly associated with biological age acceleration assessed by GrimAge acceleration (GAA) and PhenoAge acceleration (PAA) in middle-aged (45-64 years, n = 1866) and older (65-92 years, n = 1267) participants while not in young participants (24-44 years, n = 690). For example, one additional standard drink of alcohol (~14 grams of ethanol per day) was associated with a 0.71 ± 0.15-year (p = 2.1e-6) and 0.60 ± 0.18-year (p = 7.5e-4) increase in PAA in middle-aged and older participants, respectively, but the association was not significant in young participants (p = 0.23). One additional standard serving of liquor (~14 grams of ethanol) was associated with a greater increase in GAA (0.82-year, p = 4.8e-4) and PAA (1.45-year, p = 7.4e-5) than beer (GAA: 0.45-year, p = 5.2e-4; PAA: 0.48-year, p = 0.02) and wine (GAA: 0.51-year, p = 0.02; PAA: 0.91-year, p = 0.008) in middle-aged participant group. We observed that up to 28% of the association between alcohol consumption and hypertension was mediated by GAA or PAA in the pooled sample. Our findings suggest that alcohol consumption is associated with greater biological aging quantified by epigenetic aging metrics, which may mediate the association of alcohol consumption with quantitative traits, such as hypertension.

Rare variants in long non-coding RNAs are associated with blood lipid levels in the TOPMed whole-genome sequencing study.

Long non-coding RNAs (lncRNAs) are known to perform important regulatory functions in lipid metabolism. Large-scale whole-genome sequencing (WGS) studies and new statistical methods for variant set tests now provide an opportunity to assess more associations between rare variants in lncRNA genes and complex traits across the genome. In this study, we used high-coverage WGS from 66,329 participants of diverse ancestries with measurement of blood lipids and lipoproteins (LDL-C, HDL-C, TC, and TG) in the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) program to investigate the role of lncRNAs in lipid variability. We aggregated rare variants for 165,375 lncRNA genes based on their genomic locations and conducted rare-variant aggregate association tests using the STAAR (variant-set test for association using annotation information) framework. We performed STAAR conditional analysis adjusting for common variants in known lipid GWAS loci and rare-coding variants in nearby protein-coding genes. Our analyses revealed 83 rare lncRNA variant sets significantly associated with blood lipid levels, all of which were located in known lipid GWAS loci (in a ±500-kb window of a Global Lipids Genetics Consortium index variant). Notably, 61 out of 83 signals (73%) were conditionally independent of common regulatory variation and rare protein-coding variation at the same loci. We replicated 34 out of 61 (56%) conditionally independent associations using the independent UK Biobank WGS data. Our results expand the genetic architecture of blood lipids to rare variants in lncRNAs.

Expression quantitative trait methylation analysis elucidates gene regulatory effects of DNA methylation: the Framingham Heart Study.

Expression quantitative trait methylation (eQTM) analysis identifies DNA CpG sites at which methylation is associated with gene expression. The present study describes an eQTM resource of CpG-transcript pairs derived from whole blood DNA methylation and RNA sequencing gene expression data in 2115 Framingham Heart Study participants. We identified 70,047 significant cis CpG-transcript pairs at p < 1E-7 where the top most significant eGenes (i.e., gene transcripts associated with a CpG) were enriched in biological pathways related to cell signaling, and for 1208 clinical traits (enrichment false discovery rate [FDR] ≤ 0.05). We also identified 246,667 significant trans CpG-transcript pairs at p < 1E-14 where the top most significant eGenes were enriched in biological pathways related to activation of the immune response, and for 1191 clinical traits (enrichment FDR ≤ 0.05). Independent and external replication of the top 1000 significant cis and trans CpG-transcript pairs was completed in the Women's Health Initiative and Jackson Heart Study cohorts. Using significant cis CpG-transcript pairs, we identified significant mediation of the association between CpG sites and cardiometabolic traits through gene expression and identified shared genetic regulation between CpGs and transcripts associated with cardiometabolic traits. In conclusion, we developed a robust and powerful resource of whole blood eQTM CpG-transcript pairs that can help inform future functional studies that seek to understand the molecular basis of disease.

Epigenome-wide DNA methylation association study of circulating IgE levels identifies novel targets for asthma.

BACKGROUND: Identifying novel epigenetic signatures associated with serum immunoglobulin E (IgE) may improve our understanding of molecular mechanisms underlying asthma and IgE-mediated diseases. METHODS: We performed an epigenome-wide association study using whole blood from Framingham Heart Study (FHS; n = 3,471, 46% females) participants and validated results using the Childhood Asthma Management Program (CAMP; n = 674, 39% females) and the Genetic Epidemiology of Asthma in Costa Rica Study (CRA; n = 787, 41% females). Using the closest gene to each IgE-associated CpG, we highlighted biologically plausible pathways underlying IgE regulation and analyzed the transcription patterns linked to IgE-associated CpGs (expression quantitative trait methylation loci; eQTMs). Using prior UK Biobank summary data from genome-wide association studies of asthma and allergy, we performed Mendelian randomization (MR) for causal inference testing using the IgE-associated CpGs from FHS with methylation quantitative trait loci (mQTLs) as instrumental variables. FINDINGS: We identified 490 statistically significant differentially methylated CpGs associated with IgE in FHS, of which 193 (39.3%) replicated in CAMP and CRA (FDR < 0.05). Gene ontology analysis revealed enrichment in pathways related to transcription factor binding, asthma, and other immunological processes. eQTM analysis identified 124 cis-eQTMs for 106 expressed genes (FDR < 0.05). MR in combination with drug-target analysis revealed CTSB and USP20 as putatively causal regulators of IgE levels (Bonferroni adjusted P < 7.94E-04) that can be explored as potential therapeutic targets. INTERPRETATION: By integrating eQTM and MR analyses in general and clinical asthma populations, our findings provide a deeper understanding of the multidimensional inter-relations of DNA methylation, gene expression, and IgE levels. FUNDING: US NIH/NHLBI grants: P01HL132825, K99HL159234. N01-HC-25195 and HHSN268201500001I.

Circulating Metabolites May Illustrate Relationship of Alcohol Consumption with Cardiovascular Disease.

Background: Metabolite signatures of long-term alcohol consumption are lacking. To better understand the molecular basis linking alcohol drinking and cardiovascular disease (CVD), we investigated circulating metabolites associated with long-term alcohol consumption and examined whether these metabolites were associated with incident CVD. Methods: Cumulative average alcohol consumption (g/day) was derived from the total consumption of beer, wine and liquor on average of 19 years in 2,428 Framingham Heart Study Offspring participants (mean age 56 years, 52% women). We used linear mixed models to investigate the associations of alcohol consumption with 211 log-transformed plasma metabolites, adjusting for age, sex, batch, smoking, diet, physical activity, BMI, and familial relationship. Cox models were used to test the association of alcohol-related metabolite scores with fatal and nonfatal incident CVD (myocardial infarction, coronary heart disease, stroke, and heart failure). Results: We identified 60 metabolites associated with cumulative average alcohol consumption (p<0.05/211≈0.00024). For example, one g/day increase of alcohol consumption was associated with higher levels of cholesteryl esters (e.g., CE 16:1, beta=0.023±0.002, p=6.3e-45) and phosphatidylcholine (e.g., PC 32:1, beta=0.021±0.002, p=3.1e-38). Survival analysis identified that 10 alcohol-associated metabolites were also associated with a differential CVD risk after adjusting for age, sex, and batch. Further, we built two alcohol consumption weighted metabolite scores using these 10 metabolites and showed that, with adjustment age, sex, batch, and common CVD risk factors, the two scores had comparable but opposite associations with incident CVD, hazard ratio 1.11(95% CI=[1.02, 1.21],p=0.02) vs 0.88 (95% CI=[0.78, 0.98], p=0.02). Summary: We identified 60 long-term alcohol consumption-associated metabolites. The association analysis with incident CVD suggests a complex metabolic basis between alcohol consumption and CVD.

Rare variants in long non-coding RNAs are associated with blood lipid levels in the TOPMed Whole Genome Sequencing Study.

Long non-coding RNAs (lncRNAs) are known to perform important regulatory functions. Large-scale whole genome sequencing (WGS) studies and new statistical methods for variant set tests now provide an opportunity to assess the associations between rare variants in lncRNA genes and complex traits across the genome. In this study, we used high-coverage WGS from 66,329 participants of diverse ancestries with blood lipid levels (LDL-C, HDL-C, TC, and TG) in the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) program to investigate the role of lncRNAs in lipid variability. We aggregated rare variants for 165,375 lncRNA genes based on their genomic locations and conducted rare variant aggregate association tests using the STAAR (variant-Set Test for Association using Annotation infoRmation) framework. We performed STAAR conditional analysis adjusting for common variants in known lipid GWAS loci and rare coding variants in nearby protein coding genes. Our analyses revealed 83 rare lncRNA variant sets significantly associated with blood lipid levels, all of which were located in known lipid GWAS loci (in a ±500 kb window of a Global Lipids Genetics Consortium index variant). Notably, 61 out of 83 signals (73%) were conditionally independent of common regulatory variations and rare protein coding variations at the same loci. We replicated 34 out of 61 (56%) conditionally independent associations using the independent UK Biobank WGS data. Our results expand the genetic architecture of blood lipids to rare variants in lncRNA, implicating new therapeutic opportunities.

Smoking, blood DNA methylation sites and lung cancer risk.

Altered DNA methylation (DNAm) might be a biological intermediary in the pathway from smoking to lung cancer. In this study, we investigated the contribution of differential blood DNAm to explain the association between smoking and lung cancer incidence. Blood DNAm was measured in 2321 Strong Heart Study (SHS) participants. Incident lung cancer was assessed as time to event diagnoses. We conducted mediation analysis, including validation with DNAm and paired gene expression data from the Framingham Heart Study (FHS). In the SHS, current versus never smoking and pack-years single-mediator models showed, respectively, 29 and 21 differentially methylated positions (DMPs) for lung cancer with statistically significant mediated effects (14 of 20 available, and five of 14 available, positions, replicated, respectively, in FHS). In FHS, replicated DMPs showed gene expression downregulation largely in trans, and were related to biological pathways in cancer. The multimediator model identified that DMPs annotated to the genes AHRR and IER3 jointly explained a substantial proportion of lung cancer. Thus, the association of smoking with lung cancer was partly explained by differences in baseline blood DNAm at few relevant sites. Experimental studies are needed to confirm the biological role of identified eQTMs and to evaluate potential implications for early detection and control of lung cancer.

Dietary and supplemental intake of vitamins C and E is associated with altered DNA methylation in an epigenome-wide association study meta-analysis.

BACKGROUND: Dietary intake of antioxidants such as vitamins C and E protect against oxidative stress, and may also be associated with altered DNA methylation patterns. METHODS: We meta-analysed epigenome-wide association study (EWAS) results from 11,866 participants across eight population-based cohorts to evaluate the association between self-reported dietary and supplemental intake of vitamins C and E with DNA methylation. EWAS were adjusted for age, sex, BMI, caloric intake, blood cell type proportion, smoking status, alcohol consumption, and technical covariates. Significant results of the meta-analysis were subsequently evaluated in gene set enrichment analysis (GSEA) and expression quantitative trait methylation (eQTM) analysis. RESULTS: In meta-analysis, methylation at 4,656 CpG sites was significantly associated with vitamin C intake at FDR ≤ 0.05. The most significant CpG sites associated with vitamin C (at FDR ≤ 0.01) were enriched for pathways associated with systems development and cell signalling in GSEA, and were associated with downstream expression of genes enriched in the immune response in eQTM analysis. Furthermore, methylation at 160 CpG sites was significantly associated with vitamin E intake at FDR ≤ 0.05, but GSEA and eQTM analysis of the top most significant CpG sites associated with vitamin E did not identify significant enrichment of any biological pathways investigated. CONCLUSIONS: We identified significant associations of many CpG sites with vitamin C and E intake, and our results suggest that vitamin C intake may be associated with systems development and the immune response.

Integrative Mendelian randomization reveals the soluble receptor for advanced glycation end products as protective in relation to rheumatoid arthritis.

Rheumatoid arthritis (RA) is a risk factor for atherosclerotic cardiovascular diseases (CVD). Given the critical roles of the immune system and inflammatory signals in the pathogenesis of CVD, we hypothesized that interrogation of CVD-related proteins using integrative genomics might provide new insights into the pathophysiology of RA. We utilized two-sample Mendelian randomization (MR) for causal inference between circulating protein levels and RA by incorporating genetic variants, followed by colocalization to characterize the causal associations. Genetic variants from three sources were obtained: those associated with 71 CVD-related proteins measured in nearly 7000 Framingham Heart Study participants, a published genome-wide association study (GWAS) of RA (19 234 cases, 61 565 controls), and GWAS of rheumatoid factor (RF) levels from the UK Biobank (n = 30 565). We identified the soluble receptor for advanced glycation end products (sRAGE), a critical inflammatory pathway protein, as putatively causal and protective for both RA (odds ratio per 1-standard deviation increment in inverse-rank normalized sRAGE level = 0.364; 95% confidence interval 0.342-0.385; P = 6.40 × 10-241) and RF levels (ß [change in RF level per sRAGE increment] = - 1.318; SE = 0.434; P = 0.002). Using an integrative genomic approach, we highlight the AGER/RAGE axis as a putatively causal and promising therapeutic target for RA.

Transcriptome-wide association study of circulating IgE levels identifies novel targets for asthma and allergic diseases.

Measurement of circulating immunoglobulin E (IgE) concentration is helpful for diagnosing and treating asthma and allergic diseases. Identifying gene expression signatures associated with IgE might elucidate novel pathways for IgE regulation. To this end, we performed a discovery transcriptome-wide association study to identify differentially expressed genes associated with circulating IgE levels in whole-blood derived RNA from 5,345 participants in the Framingham Heart Study across 17,873 mRNA gene-level transcripts. We identified 216 significant transcripts at a false discovery rate <0.05. We conducted replication using the meta-analysis of two independent external studies: the Childhood Asthma Management Program (n=610) and the Genetic Epidemiology of Asthma in Costa Rica Study (n=326); we then reversed the discovery and replication cohorts, which revealed 59 significant genes that replicated in both directions. Gene ontology analysis revealed that many of these genes were implicated in immune function pathways, including defense response, inflammatory response, and cytokine production. Mendelian randomization (MR) analysis revealed four genes (CLC, CCDC21, S100A13, and GCNT1) as putatively causal (p<0.05) regulators of IgE levels. GCNT1 (beta=1.5, p=0.01)-which is a top result in the MR analysis of expression in relation to asthma and allergic diseases-plays a role in regulating T helper type 1 cell homing, lymphocyte trafficking, and B cell differentiation. Our findings build upon prior knowledge of IgE regulation and provide a deeper understanding of underlying molecular mechanisms. The IgE-associated genes that we identified-particularly those implicated in MR analysis-can be explored as promising therapeutic targets for asthma and IgE-related diseases.

Obesity impacts the expression of Alzheimer's disease-related genes: The Framingham Heart Study.

INTRODUCTION: We investigated associations of obesity with the expression of Alzheimer's disease (AD)-related genes in a large community-based cohort. METHODS: The sample consisted of 5619 participants from the Framingham Heart Study. Obesity metrics included body mass index (BMI) and waist-to-hip ratio (WHR). Gene expression was measured for a set of 74 AD-related genes, derived by integrating genome-wide association study results with functional genomics data. RESULTS: Obesity metrics were associated with the expression of 21 AD-related genes. The strongest associations were observed with CLU, CD2AP, KLC3, and FCER1G. Unique associations were noted with TSPAN14, SLC24A4 for BMI, and ZSCAN21, BCKDK for WHR. After adjustment for cardiovascular risk factors, 13 associations remained significant for BMI and 8 for WHR. Dichotomous obesity metrics exhibited unique associations with EPHX2 for BMI, and with TSPAN14 for WHR. DISCUSSION: Obesity was associated with AD-related gene expression; these findings shed light on the molecular pathways linking obesity to AD.

Blood transcriptomic biomarkers of alcohol consumption and cardiovascular disease risk factors: the Framingham Heart Study.

BACKGROUND: The relations of alcohol consumption and gene expression remain to be elucidated. MATERIALS AND METHODS: We examined cross-sectional associations between alcohol consumption and whole blood derived gene expression levels and between alcohol-associated genes and obesity, hypertension, and diabetes in 5531 Framingham Heart Study (FHS) participants. RESULTS: We identified 25 alcohol-associated genes. We further showed cross-sectional associations of 16 alcohol-associated genes with obesity, nine genes with hypertension, and eight genes with diabetes at P < 0.002. For example, we observed decreased expression of PROK2 (ß = -0.0018; 95%CI: -0.0021, -0.0007; P = 6.5e - 5) and PAX5 (ß = -0.0014; 95%CI: -0.0021, -0.0007; P = 6.5e - 5) per 1 g/day increase in alcohol consumption. Consistent with our previous observation on the inverse association of alcohol consumption with obesity and positive association of alcohol consumption with hypertension, we found that PROK2 was positively associated with obesity (OR = 1.42; 95%CI: 1.17, 1.72; P = 4.5e - 4) and PAX5 was negatively associated with hypertension (OR = 0.73; 95%CI: 0.59, 0.89; P = 1.6e - 3). We also observed that alcohol consumption was positively associated with expression of ABCA13 (ß = 0.0012; 95%CI: 0.0007, 0.0017; P = 1.3e - 6) and ABCA13 was positively associated with diabetes (OR = 2.57; 95%CI: 1.73, 3.84; P = 3.5e - 06); this finding, however, was inconsistent with our observation of an inverse association between alcohol consumption and diabetes. CONCLUSIONS: We showed strong cross-sectional associations between alcohol consumption and expression levels of 25 genes in FHS participants. Nonetheless, complex relationships exist between alcohol-associated genes and CVD risk factors.

Assisted clustering of gene expression data using regulatory data from partially overlapping sets of individuals.

BACKGROUND: As omics measurements profiled on different molecular layers are interconnected, integrative approaches that incorporate the regulatory effect from multi-level omics data are needed. When the multi-level omics data are from the same individuals, gene expression (GE) clusters can be identified using information from regulators like genetic variants and DNA methylation. When the multi-level omics data are from different individuals, the choice of integration approaches is limited. METHODS: We developed an approach to improve GE clustering from microarray data by integrating regulatory data from different but partially overlapping sets of individuals. We achieve this through (1) decomposing gene expression into the regulated component and the other component that is not regulated by measured factors, (2) optimizing the clustering goodness-of-fit objective function. We do not require the availability of different omics measurements on all individuals. A certain amount of individual overlap between GE data and the regulatory data is adequate for modeling the regulation, thus improving GE clustering. RESULTS: A simulation study shows that the performance of the proposed approach depends on the strength of the GE-regulator relationship, degree of missingness, data dimensionality, sample size, and the number of clusters. Across the various simulation settings, the proposed method shows competitive performance in terms of accuracy compared to the alternative K-means clustering method, especially when the clustering structure is due mostly to the regulated component, rather than the unregulated component. We further validate the approach with an application to 8,902 Framingham Heart Study participants with data on up to 17,873 genes and regulation information of DNA methylation and genotype from different but partially overlapping sets of participants. We identify clustering structures of genes associated with pulmonary function while incorporating the predicted regulation effect from the measured regulators. We further investigate the over-representation of these GE clusters in pathways of other diseases that may be related to lung function and respiratory health. CONCLUSION: We propose a novel approach for clustering GE with the assistance of regulatory data that allowed for different but partially overlapping sets of individuals to be included in different omics data.

Whole genome DNA and RNA sequencing of whole blood elucidates the genetic architecture of gene expression underlying a wide range of diseases.

To create a scientific resource of expression quantitative trail loci (eQTL), we conducted a genome-wide association study (GWAS) using genotypes obtained from whole genome sequencing (WGS) of DNA and gene expression levels from RNA sequencing (RNA-seq) of whole blood in 2622 participants in Framingham Heart Study. We identified 6,778,286 cis-eQTL variant-gene transcript (eGene) pairs at p < 5 × 10-8 (2,855,111 unique cis-eQTL variants and 15,982 unique eGenes) and 1,469,754 trans-eQTL variant-eGene pairs at p < 1e-12 (526,056 unique trans-eQTL variants and 7233 unique eGenes). In addition, 442,379 cis-eQTL variants were associated with expression of 1518 long non-protein coding RNAs (lncRNAs). Gene Ontology (GO) analyses revealed that the top GO terms for cis-eGenes are enriched for immune functions (FDR < 0.05). The cis-eQTL variants are enriched for SNPs reported to be associated with 815 traits in prior GWAS, including cardiovascular disease risk factors. As proof of concept, we used this eQTL resource in conjunction with genetic variants from public GWAS databases in causal inference testing (e.g., COVID-19 severity). After Bonferroni correction, Mendelian randomization analyses identified putative causal associations of 60 eGenes with systolic blood pressure, 13 genes with coronary artery disease, and seven genes with COVID-19 severity. This study created a comprehensive eQTL resource via BioData Catalyst that will be made available to the scientific community. This will advance understanding of the genetic architecture of gene expression underlying a wide range of diseases.

Elucidating the genetic architecture of DNA methylation to identify promising molecular mechanisms of disease.

DNA methylation commonly occurs at cytosine-phosphate-guanine sites (CpGs) that can serve as biomarkers for many diseases. We analyzed whole genome sequencing data to identify DNA methylation quantitative trait loci (mQTLs) in 4126 Framingham Heart Study participants. Our mQTL mapping identified 94,362,817 cis-mQTLvariant-CpG pairs (for 210,156 unique autosomal CpGs) at P < 1e-7 and 33,572,145 trans-mQTL variant-CpG pairs (for 213,606 unique autosomal CpGs) at P < 1e-14. Using cis-mQTL variants for 1258 CpGs associated with seven cardiovascular disease (CVD) risk factors, we found 104 unique CpGs that colocalized with at least one CVD trait. For example, cg11554650 (PPP1R18) colocalized with type 2 diabetes, and was driven by a single nucleotide polymorphism (rs2516396). We performed Mendelian randomization (MR) analysis and demonstrated 58 putatively causal relations of CVD risk factor-associated CpGs to one or more risk factors (e.g., cg05337441 [APOB] with LDL; MR P = 1.2e-99, and 17 causal associations with coronary artery disease (e.g. cg08129017 [SREBF1] with coronary artery disease; MR P = 5e-13). We also showed that three CpGs, e.g., cg14893161 (PM20D1), are putatively causally associated with COVID-19 severity. To assist in future analyses of the role of DNA methylation in disease pathogenesis, we have posted a comprehensive summary data set in the National Heart, Lung, and Blood Institute's BioData Catalyst.

Whole Genome DNA and RNA Sequencing of Whole Blood Elucidates the Genetic Architecture of Gene Expression Underlying a Wide Range of Diseases.

To create a scientific resource of expression quantitative trail loci (eQTL), we conducted a genome-wide association study (GWAS) using genotypes obtained from whole genome sequencing (WGS) of DNA and gene expression levels from RNA sequencing (RNA-seq) of whole blood in 2622 participants in Framingham Heart Study. We identified 6,778,286 cis -eQTL variant-gene transcript (eGene) pairs at p < 5x10 - 8 (2,855,111 unique cis -eQTL variants and 15,982 unique eGenes) and 1,469,754 trans -eQTL variant-eGene pairs at p < 1e-12 (526,056 unique trans -eQTL variants and 7,233 unique eGenes). In addition, 442,379 cis -eQTL variants were associated with expression of 1518 long non-protein coding RNAs (lncRNAs). Gene Ontology (GO) analyses revealed that the top GO terms for cis- eGenes are enriched for immune functions (FDR < 0.05). The cis -eQTL variants are enriched for SNPs reported to be associated with 815 traits in prior GWAS, including cardiovascular disease risk factors. As proof of concept, we used this eQTL resource in conjunction with genetic variants from public GWAS databases in causal inference testing (e.g., COVID-19 severity). After Bonferroni correction, Mendelian randomization analyses identified putative causal associations of 60 eGenes with systolic blood pressure, 13 genes with coronary artery disease, and seven genes with COVID-19 severity. This study created a comprehensive eQTL resource via BioData Catalyst that will be made available to the scientific community. This will advance understanding of the genetic architecture of gene expression underlying a wide range of diseases.