Dataset Information

Selecting likely causal risk factors from high-throughput experiments using multivariable Mendelian randomization.

ABSTRACT: Modern high-throughput experiments provide a rich resource to investigate causal determinants of disease risk. Mendelian randomization (MR) is the use of genetic variants as instrumental variables to infer the causal effect of a specific risk factor on an outcome. Multivariable MR is an extension of the standard MR framework to consider multiple potential risk factors in a single model. However, current implementations of multivariable MR use standard linear regression and hence perform poorly with many risk factors. Here, we propose a two-sample multivariable MR approach based on Bayesian model averaging (MR-BMA) that scales to high-throughput experiments. In a realistic simulation study, we show that MR-BMA can detect true causal risk factors even when the candidate risk factors are highly correlated. We illustrate MR-BMA by analysing publicly-available summarized data on metabolites to prioritise likely causal biomarkers for age-related macular degeneration.

SUBMITTER: Zuber V

PROVIDER: S-EPMC6946691 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Selecting likely causal risk factors from high-throughput experiments using multivariable Mendelian randomization.

Zuber Verena V Colijn Johanna Maria JM Klaver Caroline C Burgess Stephen S

Nature communications 20200107 1

Modern high-throughput experiments provide a rich resource to investigate causal determinants of disease risk. Mendelian randomization (MR) is the use of genetic variants as instrumental variables to infer the causal effect of a specific risk factor on an outcome. Multivariable MR is an extension of the standard MR framework to consider multiple potential risk factors in a single model. However, current implementations of multivariable MR use standard linear regression and hence perform poorly w ...[more]

PMID: 31911605

Similar Datasets

Project description:BackgroundObesity (waist circumference, body mass index (BMI)) and lifestyle factors (dietary habits, smoking, alcohol drinking, Sedentary behavior) have been associated with risk of benign prostatic hyperplasia (BPH) in observational studies, but whether these associations are causal is unclear.MethodsWe performed a univariable and multivariable Mendelian randomization study to evaluate these associations. Genetic instruments associated with exposures at the genome-wide significance level (P < 5 × 10-8) were selected from corresponding genome-wide associations studies (n = 216,590 to 1,232,091 individuals). Summary-level data for BPH were obtained from the UK Biobank (14,126 cases and 169,762 non-cases) and FinnGen consortium (13,118 cases and 72,799 non-cases). Results from UK Biobank and FinnGen consortium were combined using fixed-effect meta-analysis.ResultsThe combined odds ratios (ORs) of BPH were 1.24 (95% confidence interval (CI), 1.07-1.43, P = 0.0045), 1.08 (95% CI 1.01-1.17, P = 0.0175), 0.94 (95% CI 0.67-1.30, P = 0.6891), 1.29 (95% CI 0.88-1.89, P = 0.1922), 1.23 (95% CI 0.85-1.78, P = 0.2623), and 1.04 (95% CI 0.76-1.42, P = 0.8165) for one standard deviation (SD) increase in waist circumference, BMI, and relative carbohydrate, fat, protein and sugar intake, 1.05 (95% CI 0.92-1.20, P = 0.4581) for one SD increase in prevalence of smoking initiation, 1.10 (95% CI 0.96-1.26, P = 0.1725) and 0.84 (95% CI 0.69-1.02, P = 0.0741) for one SD increase of log-transformed smoking per day and drinks per week, and 1.31 (95% CI 1.08-1.58, P = 0.0051) for one SD increase in sedentary behavior. Genetically predicted waist circumference (OR = 1.26, 95% CI 1.11-1.43, P = 0.0004) and sedentary behavior (OR = 1.14, 95% CI 1.05-1.23, P = 0.0021) were associated with BPH after the adjustment of BMI.ConclusionThis study supports independent causal roles of high waist circumference, BMI and sedentary behavior in BPH.

Project description:IntroductionThe correlation between dyslipidemia and periodontitis is revealed through epidemiological studies. However, the results are affected by several confounding factors. This study aims to elucidate the genetic causal association between circulating lipid traits and periodontitis by two-sample Mendelian randomization (MR) analysis.MethodsAfter the different screening processes, two cohorts of circulating lipid traits from the UK Biobank were used as exposure data, including five circulating lipid traits. The Periodontitis cohort was selected from the GeneLifestyle Interactions in Dental Endpoints (GLIDE) consortium as outcome data. In univariable MR, the inverse variance weighted (IVW) was used in conjunction with six additional analytical methods to assess causality. The Cochran Q test, IGX 2 statistic, MR-PRESSO, and MR-Egger intercept were used to quantify heterogeneity and pleiotropy. The multivariable MR-IVW (MVMR-IVW) and MVMR-robust were mainly used as analytical methods in the multiple MR analyses.ResultsThe IVW estimates showed that genetically predicted Apolipoprotein A1 (apo A1) [odds ratio (OR)=1.158, 95% confidence interval (CI)=1.007-1.331, P-value=0.040] was potentially associated with the risk of periodontitis, but the statistical power of the results was low. Multivariable MR analysis did not reveal any significant causal relationship between apo A1 and periodontitis (OR=0.72, 95% CI=0.36-1.41, P-value=0.34). In the validation cohort, there was also no significant causal relationship between apo A1 and periodontitis (OR=1.079, 95% CI=0.903-1.290, P-value=0.401). Meanwhile, genetically predicted Apolipoprotein B (apo B), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglyceride (TG) (all P-values>0.05) were not significantly associated with the risk of periodontitis causal inference.ConclusionThis MR analysis was unable to provide genetic evidence for the influence of these five circulating lipid traits on periodontitis. However, a more extensive study with a more comprehensive circulating lipid profile and periodontitis data is needed due to study limitations.

Dataset Information

Selecting likely causal risk factors from high-throughput experiments using multivariable Mendelian randomization.

Publications

Selecting likely causal risk factors from high-throughput experiments using multivariable Mendelian randomization.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets