Causal modeling in a multi-omic setting: insights from GAW20

Background

The GAW20 group formed on the theme of methods for association analyses of repeated measures comprised 4sets of investigators. The provided “real” data set included genotypes obtained from a human whole-genome association study based on longitudinal measurements of triglycerides (TGs) and high-density lipoprotein in addition to methylation levels before and after administration of fenofibrate. The simulated data set contained 200 replications of methylation levels and posttreatment TGs, mimicking the real data set.

Results

The different investigators in the group focused on the statistical challenges unique to family-based association analyses of phenotypes measured longitudinally and applied a wide spectrum of statistical methods such as linear mixed models, generalized estimating equations, and quasi-likelihood–based regression models. This article discusses the varying strategies explored by the group’s investigators with the common goal of improving the power to detect association with repeated measures of a phenotype.

Conclusions

Although it is difficult to identify a common message emanating from the different contributions because of the diversity in the issues addressed, the unifying theme of the contributions lie in the search for novel analytic strategies to circumvent the limitations of existing methodologies to detect genetic association.

     

Identification of a new locus and validation of previously reported loci showing differential methylation associated with smoking. The REGICOR study

Smoking increases the risk of many diseases and could act through changes in DNA methylation patterns. The aims of this study were to determine the association between smoking and DNA methylation throughout the genome at cytosine-phosphate-guanine (CpG) site level and genomic regions. A discovery cross-sectional epigenome-wide association study nested in the follow-up of the REGICOR cohort was designed and included 645 individuals. Blood DNA methylation was assessed using the Illumina HumanMethylation450 BeadChip. Smoking status was self-reported using a standardized questionnaire. We identified 66 differentially methylated CpG sites associated with smoking, located in 38 genes. In most of these CpG sites, we observed a trend among those quitting smoking to recover methylation levels typical of never smokers. A CpG site located in a novel smoking-associated gene (cg06394460 in LNX2) was hypomethylated in current smokers. Moreover, we validated two previously reported CpG sites (cg05886626 in THBS1, and cg24838345 in MTSS1) for their potential relation to atherosclerosis and cancer diseases, using several different approaches: CpG site methylation, gene expression, and plasma protein level determinations. Smoking was also associated with higher THBS1 gene expression but with lower levels of thrombospondin-1 in plasma. Finally, we identified differential methylation regions in 13 genes and in four non-coding RNAs. In summary, this study replicated previous findings and identified and validated a new CpG site located in LNX2 associated with smoking.     

Causal modeling in a multi-omic setting: insights from GAW20

Background

Increasingly available multilayered omics data on large populations has opened exciting analytic opportunities and posed unique challenges to robust estimation of causal effects in the setting of complex disease phenotypes. The GAW20 Causal Modeling Working Group has applied complementary approaches (eg, Mendelian randomization, structural equations modeling, Bayesian networks) to discover novel causal effects of genomic and epigenomic variation on lipid phenotypes, as well as to validate prior findings from observational studies.

Results

Two Mendelian randomization studies have applied novel approaches to instrumental variable selection in methylation data, identifying bidirectional causal effects of CPT1A and triglycerides, as well as of RNMT and C6orf42, on high-density lipoprotein cholesterol response to fenofibrate. The CPT1A finding also emerged in a Bayesian network study. The Mendelian randomization studies have implemented both existing and novel steps to account for pleiotropic effects, which were independently detected in the GAW20 data via a structural equation modeling approach. Two studies estimated indirect effects of genomic variation (via DNA methylation and/or correlated phenotypes) on lipid outcomes of interest. Finally, a novel weighted R² measure was proposed to complement other causal inference efforts by controlling for the influence of outlying observations.

Conclusions

The GAW20 contributions illustrate the diversity of possible approaches to causal inference in the multi-omic context, highlighting the promises and assumptions of each method and the benefits of integrating both across methods and across omics layers for the most robust and comprehensive insights into disease processes.

     

DNA methylation and obesity traits: An epigenome-wide association study. The REGICOR study

Obesity is associated with increased risk of several diseases and has become epidemic. Obesity is highly heritable but the genetic variants identified by genome-wide association studies explain only limited variability. Epigenetics could contribute to explain the missing variability. The study aim was to discover differential methylation patterns related to obesity. We designed an epigenome-wide association study with a discovery phase in a subsample of 641 REGICOR study participants, validated by analysis of 2,515 participants in the Framingham Offspring Study. Blood DNA methylation was assessed using Illumina HumanMethylation450 BeadChip. Next, we meta-analyzed the data using the fixed effects method and performed a functional and pathway analysis using the Ingenuity Pathway Analysis software. We were able to validate 94 CpGs associated with body mass index (BMI) and 49 CpGs associated with waist circumference, located in 95 loci. In addition, we newly discovered 70 CpGs associated with BMI and 33 CpGs related to waist circumference. These CpGs explained 25.94% and 29.22% of the variability of BMI and waist circumference, respectively, in the REGICOR sample. We also evaluated 65 of the 95 validated loci in the GIANT genome-wide association data; 10 of them had Tag SNPs associated with BMI. The top-ranked diseases and functions identified in the functional and pathway analysis were neurologic, psychological, endocrine, and metabolic.