MR-Clust: Clustering of genetic variants in Mendelian randomization with similar causal estimates

Kirk, Paul; Burgess, Stephen

doi:10.1101/2019.12.18.881326

Cited by 11 publications

(15 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have also highlighted that there is an important heterogeneity in causal effect estimates that vary with the IV selection threshold, due to heterogeneity in the estimates between the groups of genetic variants used for different thresholds. This can happen if there is a strong phenotypic heterogeneity in the exposure, in which case different groups of IVs could be affecting the exposure through different pathways [21] . Alternatively, in the presence of a genetic confounder, IVs picked up at a less stringent thresholds may be associated to the confounder, hence violating the second assumption of MR.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Bias correction for inverse variance weighting Mendelian randomization

Mounier

Kutalik

2021

Preprint

View full text Add to dashboard Cite

Inverse-variance weighted two-sample Mendelian Randomization (IVW-MR) is the most widely used approach that uses genome-wide association studies summary statistics to infer the existence and strength of the causal effect between an exposure and an outcome. Estimates from this approach can be subject to different biases due to: (i) the overlap between the exposure and outcome samples; (ii) the use of weak instruments and winner's curse. We developed a method that aims at tackling all these biases together. Assuming spike-and-slab genomic architecture and leveraging LD-score regression and other techniques, we could analytically derive and reliably estimate the bias of IVW-MR using association summary statistics only. This allowed us to apply a bias correction to IVW-MR estimates, which we tested using simulated data for a wide range of realistic scenarios. In all the explored scenarios, our correction reduced the bias, in some situations by as much as 30 folds. When applied to real data on obesity-related exposures, we observed significant differences between IVW-based and corrected effects, both for non-overlapping and fully overlapping samples. While most studies are extremely careful to avoid any sample overlap when performing two-sample MR analysis, we have demonstrated that the incurred bias is much less substantial than the one due to weak instruments or winner's curse, which are often ignored.

show abstract

Section: Discussionmentioning

confidence: 99%

“…Such phenomenon is out of the scope of our paper. In such case, IVW two-sample MR estimates would be biased, and more sophisticated approaches either specifically accounting for this genetic confounding (CAUSE [22] , LHC-MR [23] ) or others allowing for multiple causal effects (MR-Clust [21] ) would be needed.…”

Section: Application To Ukbbmentioning

confidence: 99%

Bias correction for inverse variance weighting Mendelian randomization

Mounier

Kutalik

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We therefore examined a scenario where variants can be divided into different clusters. According to MR-Clust, each IV is only assigned to a cluster if the conditional probability of belonging to that cluster is high (larger than 0.8) and clusters are only displayed if at least four IVs are assigned to it (Foley et al, 2019). As shown in Figure 2, for IBD, we observed two distinct clusters suggesting one strong positive causal effect and one strong negative causal effect; for SLE, we observed a single cluster suggesting a strong positive causal effect; and for MS, we observed a single cluster suggesting a strong negative causal effect.…”

Section: Resultsmentioning

confidence: 99%

“…Mendelian randomization methods evaluate an overall casual estimation; it is likely that several distinct causal mechanisms underlie the alcohol-disease relationship, in which a risk factor influences outcome with different magnitudes of causal effect. We examined such a scenario through MR-Clust (Foley et al, 2019), an approach that divides IVs into distinct clusters such that all variants in the cluster have similar causal estimates.…”

Section: Mendelian Randomization Analysismentioning

confidence: 99%

Alcohol Consumption and Risk of Common Autoimmune Inflammatory Diseases—Evidence From a Large-Scale Genetic Analysis Totaling 1 Million Individuals

et al. 2021

View full text Add to dashboard Cite

Purpose: Observational studies have suggested a protective effect of alcohol intake with autoimmune disorders, which was not supported by Mendelian randomization (MR) analyses that used only a few (<20) instrumental variables.Methods: We systemically interrogated a putative causal relationship between alcohol consumption and four common autoimmune disorders, using summary-level data from the largest genome-wide association study (GWAS) conducted on inflammatory bowel disease (IBD), rheumatoid arthritis (RA), multiple sclerosis (MS), and systemic lupus erythematosus (SLE). We quantified the genetic correlation to examine a shared genetic similarity. We constructed a strong instrument using 99 genetic variants associated with drinks per week and applied several two-sample MR methods. We additionally incorporated excessive drinking as reflected by alcohol use disorder identification test score.Results: We observed a negatively shared genetic basis between alcohol intake and autoimmune disorders, although none was significant (rg = −0.07 to −0.02). For most disorders, genetically predicted alcohol consumption was associated with a slightly (10–25%) decreased risk of onset, yet these associations were not significant. Meta-analyzing across RA, MS, and IBD, the three Th1-related disorders yielded to a marginally significantly reduced effect [OR = 0.70 (0.51–0.95), P = 0.02]. Excessive drinking did not appear to reduce the risk of autoimmune disorders.Conclusions: With its greatly augmented sample size and substantially improved statistical power, our MR study does not convincingly support a beneficial role of alcohol consumption in each individual autoimmune disorder. Future studies may be designed to replicate our findings and to understand a causal effect on disease prognosis.

show abstract

“…We further examined if different aspects of obesity play an aetiological role in different reproductive conditions. For each obesity trait-reproductive disease pair, we grouped the genetic instruments for the obesity traits by those that do not have an effect on the disease (“null cluster”), those which have a similar scaled effect on the disease (the “substantial clusters”), and those that have a scaled effect that cannot be grouped with other variants (“junk cluster”) using MRClust (57). One substantial cluster was identified for each pair of obesity traits and reproductive conditions.…”

Section: Resultsmentioning

confidence: 99%

The role of obesity in female reproductive conditions: A Mendelian Randomisation study

Venkatesh

Ferreira

Benónísdóttir³

et al. 2021

Preprint

View full text Add to dashboard Cite

Background: Obesity is observationally associated with altered risk of many female reproductive conditions. These include polycystic ovary syndrome (PCOS), abnormal uterine bleeding, endometriosis, infertility, and pregnancy-related disorders. However, the roles and mechanisms of obesity in the aetiology of reproductive disorders remain unclear. Methods and Findings: We estimated observational and genetically predicted causal associations between obesity, metabolic hormones, and female reproductive conditions using logistic regression, generalised additive models, and Mendelian randomisation (two-sample, non-linear, and multivariable) applied to data from UK Biobank and publicly available genome-wide association studies (GWAS). Body mass index (BMI), waist-hip ratio (WHR), and WHR adjusted for BMI (WHRadjBMI) were observationally (odds ratios (ORs) = 1.02 - 1.87 per 1 S.D. obesity trait) and causally (ORs = 1.06 - 2.09) associated with uterine fibroids (UF), PCOS, heavy menstrual bleeding (HMB), and pre-eclampsia. Causal effect estimates of WHR and WHRadjBMI, but not BMI, were attenuated compared to their observational counterparts. Genetically predicted visceral adipose tissue mass was causal for the development of HMB, PCOS, and pre-eclampsia (ORs = 1.01 - 3.38). Increased waist circumference also posed a higher causal risk (ORs = 1.16 - 1.93) for the development of these disorders and UF than did increased hip circumference (ORs = 1.06 - 1.10). Leptin, fasting insulin, and insulin resistance each mediated between 20% - 50% of the total causal effect of obesity on pre-eclampsia. Reproductive conditions clustered based on shared genetic components of their aetiological relationships with obesity. Conclusions: In this first systematic, large-scale, genetics-based analysis of the aetiological relationships between obesity and female reproductive conditions, we found that common indices of overall and central obesity increased risk of reproductive disorders to heterogenous extents, mediated by metabolic hormones. Our results suggest exploring the mechanisms mediating the causal effects of overweight and obesity on gynaecological health to identify targets for disease prevention and treatment.

show abstract

MR-Clust: Clustering of genetic variants in Mendelian randomization with similar causal estimates

Cited by 11 publications

References 30 publications

Bias correction for inverse variance weighting Mendelian randomization

Bias correction for inverse variance weighting Mendelian randomization

Alcohol Consumption and Risk of Common Autoimmune Inflammatory Diseases—Evidence From a Large-Scale Genetic Analysis Totaling 1 Million Individuals

The role of obesity in female reproductive conditions: A Mendelian Randomisation study

Contact Info

Product

Resources

About