Inference and visualization of phenome-wide causal relationships using genetic data: an application to dental caries and periodontitis

AbstractBackgroundHypothesis-free Mendelian randomization studies provide a way to assess the causal relevance of a trait across the human phenome but can be limited by statistical power or complicated by horizontal pleiotropy. The recently described latent causal variable (LCV) approach provides an alternative method for causal inference which might be useful in hypothesis-free experiments.MethodsWe developed an automated pipeline for phenome-wide tests using the LCV approach including steps to estimate partial genetic causality, filter to a meaningful set of estimates, apply correction for multiple testing and then present the findings in a graphical summary termed a causal architecture plot. We apply this process to body mass index and lipid traits as exemplars of traits where there is strong prior expectation for causal effects and dental caries and periodontitis as exemplars of traits where there is a need for causal inference.ResultsThe results for lipids and BMI suggest that these traits are best viewed as creating consequences on a multitude of traits and conditions, thus providing additional evidence that supports viewing these traits as targets for interventions to improve health. On the other hand, caries and periodontitis are best viewed as a downstream consequence of other traits and diseases rather than a cause of ill health.ConclusionsThe automated process is available as part of the MASSIVE pipeline from the Complex-Traits Genetics Virtual Lab (https://vl.genoma.io) and results are available in (https://view.genoma.io). We propose causal architecture plots based on phenome-wide partial genetic causality estimates as a way visualizing the overall causal map of the human phenome.Key messagesThe latent causal variable approach uses summary statistics from genome-wide association studies to estimate a parameter termed genetic causality proportion.Systematic estimation of genetic causality proportion for many pairs of traits provides an alternative method for phenome-wide causal inference with some theoretical and practical advantages compared to phenome-wide Mendelian randomization.Using this approach, we confirm that lipid traits are an upstream risk factor for other traits and diseases, and we identify that dental diseases are predominantly a downstream consequence of other traits rather than a cause of poor systemic health.The method assumes no bidirectional causality and no confounding by environmental correlates of genotypes, so care is needed when these assumptions are not met.We developed an automated and accessible pipeline for estimating phenome-wide causal relationships and generating interactive visual summaries.

Download Full-text

Are there causal relationships between ADHD and BMI? Evidence from multiple genetically informed designs

10.1101/2020.04.16.20067918 ◽

2020 ◽

Author(s):

Chao-Yu Liu ◽

Tabea Schoeler ◽

Neil M Davies ◽

Hugo Peyre ◽

Kai-Xiang Lim ◽

...

Keyword(s):

Mendelian Randomization ◽

Association Studies ◽

Monozygotic Twins ◽

Causal Effects ◽

List Type ◽

Genome Wide Association Studies ◽

Causal Relationships ◽

Adhd Symptoms ◽

Multiple Sources ◽

Polygenic Score

AbstractBackgroundAttention-deficit/hyperactivity disorder (ADHD) and Body Mass Index (BMI) are associated. However, it remains unclear whether this association reflects causal relationships in either direction, or confounding. Here, we implemented genetically informed methods to examine bidirectional causality and potential confounding.MethodsThree genetically informed methods were employed: (1) cross-lagged twin-differences analyses to assess bidirectional effects of ADHD symptoms and BMI at ages 8, 12, 14 and 16 years in 2386 pairs of monozygotic twins from the Twins Early Development Study (TEDS), (2) within- and between-family ADHD and BMI polygenic score (PS) analyses in 3320 pairs of dizygotic TEDS twins and (3) two-sample bidirectional Mendelian randomization (MR) using summary statistics from Genome-Wide Association Studies (GWAS) on ADHD (N=55,374) and BMI (N=806,834).ResultsMixed results were obtained across the three methods. Twin-difference analyses provided little support for cross-lagged associations between ADHD symptoms and BMI over time. PS analyses were consistent with bidirectional relationships between ADHD and BMI with plausible time-varying effects from childhood to adolescence. MR findings were also consistent with bidirectional causal effects between ADHD and BMI. Multivariable MR suggested the presence of substantial confounding in bidirectional relationships.ConclusionsThe three methods converged to highlight multiple sources of confounding in the association between ADHD and BMI. PS and MR analyses suggested plausible causal relationships in both directions. Possible explanations for mixed causal findings across methods are discussed.Key messagesPolygenic score and Mendelian randomization analyses were consistent with bidirectional causal effects between ADHD and BMI.Findings from different genetically informed methods suggested that multiple sources of confounding are at play, including genetic and shared environmental confounding, population stratification, assortative mating and dynastic effects.The ADHD polygenic score increasingly associated with BMI phenotype from childhood to adolescence, suggesting an increasing role of ADHD in the aetiology of BMI. Findings were reversed between the BMI polygenic score and ADHD.Addressing mixed evidence will require increased sample sizes to implement novel methods such as within-family MR.

Download Full-text

Causal Inference Between Chronic Periodontitis and Chronic Kidney Disease: A Bidirectional Mendelian Randomization Analysis in a European Population

Frontiers in Genetics ◽

10.3389/fgene.2021.676136 ◽

2021 ◽

Vol 12 ◽

Author(s):

Jie Yang ◽

Tianyi Chen ◽

Yahong Zhu ◽

Mingxia Bai ◽

Xingang Li

Keyword(s):

Chronic Kidney Disease ◽

Kidney Disease ◽

Causal Inference ◽

Causal Relationship ◽

Instrumental Variables ◽

Chronic Periodontitis ◽

Mendelian Randomization ◽

Association Studies ◽

Genome Wide Association Studies ◽

Summary Statistics

BackgroundPrevious epidemiological studies have shown significant associations between chronic periodontitis (CP) and chronic kidney disease (CKD), but the causal relationship remains uncertain. Aiming to examine the causal relationship between these two diseases, we conducted a bidirectional two-sample Mendelian randomization (MR) analysis with multiple MR methods.MethodsFor the casual effect of CP on CKD, we selected seven single-nucleotide polymorphisms (SNPs) specific to CP as genetic instrumental variables from the genome-wide association studies (GWAS) in the GLIDE Consortium. The summary statistics of complementary kidney function measures, i.e., estimated glomerular filtration rate (eGFR) and blood urea nitrogen (BUN), were derived from the GWAS in the CKDGen Consortium. For the reversed causal inference, six SNPs associated with eGFR and nine with BUN from the CKDGen Consortium were included and the summary statistics were extracted from the CLIDE Consortium.ResultsNo significant causal association between genetically determined CP and eGFR or BUN was found (all p > 0.05). Based on the conventional inverse variance-weighted method, one of seven instrumental variables supported genetically predicted CP being associated with a higher risk of eGFR (estimate = 0.019, 95% CI: 0.012–0.026, p < 0.001).ConclusionEvidence from our bidirectional causal inference does not support a causal relation between CP and CKD risk and therefore suggests that associations reported by previous observational studies may represent confounding.

Download Full-text

MR-LDP: a two-sample Mendelian randomization for GWAS summary statistics accounting for linkage disequilibrium and horizontal pleiotropy

NAR Genomics and Bioinformatics ◽

10.1093/nargab/lqaa028 ◽

2020 ◽

Vol 2 (2) ◽

Cited By ~ 2

Author(s):

Qing Cheng ◽

Yi Yang ◽

Xingjie Shi ◽

Kar-Fu Yeung ◽

Can Yang ◽

...

Keyword(s):

Risk Factors ◽

Linkage Disequilibrium ◽

Genetic Variants ◽

Mendelian Randomization ◽

Association Studies ◽

Alternative Methods ◽

Genome Wide Association Studies ◽

Summary Statistics ◽

Causal Relationships ◽

Disease Outcomes

Abstract The proliferation of genome-wide association studies (GWAS) has prompted the use of two-sample Mendelian randomization (MR) with genetic variants as instrumental variables (IVs) for drawing reliable causal relationships between health risk factors and disease outcomes. However, the unique features of GWAS demand that MR methods account for both linkage disequilibrium (LD) and ubiquitously existing horizontal pleiotropy among complex traits, which is the phenomenon wherein a variant affects the outcome through mechanisms other than exclusively through the exposure. Therefore, statistical methods that fail to consider LD and horizontal pleiotropy can lead to biased estimates and false-positive causal relationships. To overcome these limitations, we proposed a probabilistic model for MR analysis in identifying the causal effects between risk factors and disease outcomes using GWAS summary statistics in the presence of LD and to properly account for horizontal pleiotropy among genetic variants (MR-LDP) and develop a computationally efficient algorithm to make the causal inference. We then conducted comprehensive simulation studies to demonstrate the advantages of MR-LDP over the existing methods. Moreover, we used two real exposure–outcome pairs to validate the results from MR-LDP compared with alternative methods, showing that our method is more efficient in using all-instrumental variants in LD. By further applying MR-LDP to lipid traits and body mass index (BMI) as risk factors for complex diseases, we identified multiple pairs of significant causal relationships, including a protective effect of high-density lipoprotein cholesterol on peripheral vascular disease and a positive causal effect of BMI on hemorrhoids.

Download Full-text

Bayesian weighted Mendelian randomization for causal inference based on summary statistics

Bioinformatics ◽

10.1093/bioinformatics/btz749 ◽

2019 ◽

Author(s):

Jia Zhao ◽

Jingsi Ming ◽

Xianghong Hu ◽

Gang Chen ◽

Jin Liu ◽

...

Keyword(s):

Causal Inference ◽

Complex Traits ◽

Mendelian Randomization ◽

Causal Effect ◽

Association Studies ◽

Supplementary Information ◽

Genome Wide Association Studies ◽

Simulation Studies ◽

Genome Wide ◽

Complex Human Traits

Abstract Motivation The results from Genome-Wide Association Studies (GWAS) on thousands of phenotypes provide an unprecedented opportunity to infer the causal effect of one phenotype (exposure) on another (outcome). Mendelian randomization (MR), an instrumental variable (IV) method, has been introduced for causal inference using GWAS data. Due to the polygenic architecture of complex traits/diseases and the ubiquity of pleiotropy, however, MR has many unique challenges compared to conventional IV methods. Results We propose a Bayesian weighted Mendelian randomization (BWMR) for causal inference to address these challenges. In our BWMR model, the uncertainty of weak effects owing to polygenicity has been taken into account and the violation of IV assumption due to pleiotropy has been addressed through outlier detection by Bayesian weighting. To make the causal inference based on BWMR computationally stable and efficient, we developed a variational expectation-maximization (VEM) algorithm. Moreover, we have also derived an exact closed-form formula to correct the posterior covariance which is often underestimated in variational inference. Through comprehensive simulation studies, we evaluated the performance of BWMR, demonstrating the advantage of BWMR over its competitors. Then we applied BWMR to make causal inference between 130 metabolites and 93 complex human traits, uncovering novel causal relationship between exposure and outcome traits. Availability and implementation The BWMR software is available at https://github.com/jiazhao97/BWMR. Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

An iterative approach to detect pleiotropy and perform Mendelian Randomization analysis using GWAS summary statistics

Bioinformatics ◽

10.1093/bioinformatics/btaa985 ◽

2020 ◽

Author(s):

Xiaofeng Zhu ◽

Xiaoyin Li ◽

Rong Xu ◽

Tao Wang

Keyword(s):

Complex Traits ◽

Mendelian Randomization ◽

Causal Effect ◽

Association Studies ◽

Real Data ◽

Supplementary Information ◽

Genome Wide Association Studies ◽

Summary Statistics ◽

Causal Relationships ◽

Multiple Traits

Abstract Motivation The overall association evidence of a genetic variant with multiple traits can be evaluated by cross-phenotype association analysis using summary statistics from genome-wide association studies. Further dissecting the association pathways from a variant to multiple traits is important to understand the biological causal relationships among complex traits. Results Here, we introduce a flexible and computationally efficient Iterative Mendelian Randomization and Pleiotropy (IMRP) approach to simultaneously search for horizontal pleiotropic variants and estimate causal effect. Extensive simulations and real data applications suggest that IMRP has similar or better performance than existing Mendelian Randomization methods for both causal effect estimation and pleiotropic variant detection. The developed pleiotropy test is further extended to detect colocalization for multiple variants at a locus. IMRP will greatly facilitate our understanding of causal relationships underlying complex traits, in particular, when a large number of genetic instrumental variables are used for evaluating multiple traits. Availability and implementation The software IMRP is available at https://github.com/XiaofengZhuCase/IMRP. The simulation codes can be downloaded at http://hal.case.edu/∼xxz10/zhu-web/ under the link: MR Simulations software. Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

Association of genetically predicted blood sucrose with coronary heart disease and its risk factors in Mendelian randomization

Scientific Reports ◽

10.1038/s41598-020-78685-5 ◽

2020 ◽

Vol 10 (1) ◽

Author(s):

Ting Zhang ◽

Shiu Lun Au Yeung ◽

C. Mary Schooling

Keyword(s):

Risk Factors ◽

Coronary Heart Disease ◽

Heart Disease ◽

Dental Caries ◽

Multiple Testing ◽

Mendelian Randomization ◽

Association Studies ◽

Sensitivity Analyses ◽

Genome Wide Association Studies ◽

Standard Deviation Increase

AbstractWe assessed the associations of genetically instrumented blood sucrose with risk of coronary heart disease (CHD) and its risk factors (i.e., type 2 diabetes, adiposity, blood pressure, lipids, and glycaemic traits), using two-sample Mendelian randomization. We used blood fructose as a validation exposure. Dental caries was a positive control outcome. We selected genetic variants strongly (P < 5 × 10–6) associated with blood sucrose or fructose as instrumental variables and applied them to summary statistics from the largest available genome-wide association studies of the outcomes. Inverse-variance weighting was used as main analysis. Sensitivity analyses included weighted median, MR-Egger and MR-PRESSO. Genetically higher blood sucrose was positively associated with the control outcome, dental caries (odds ratio [OR] 1.04 per log10 transformed effect size [median-normalized standard deviation] increase, 95% confidence interval [CI] 1.002–1.08, P = 0.04), but this association did not withstand allowing for multiple testing. The estimate for blood fructose was in the same direction. Genetically instrumented blood sucrose was not clearly associated with CHD (OR 1.01, 95% CI 0.997–1.02, P = 0.14), nor with its risk factors. Findings were similar for blood fructose. Our study found some evidence of the expected detrimental effect of sucrose on dental caries but no effect on CHD. Given a small effect on CHD cannot be excluded, further investigation with stronger genetic predictors is required.

Download Full-text

Automating Mendelian randomization through machine learning to construct a putative causal map of the human phenome

10.1101/173682 ◽

2017 ◽

Cited By ~ 32

Author(s):

Gibran Hemani ◽

Jack Bowden ◽

Philip Haycock ◽

Jie Zheng ◽

Oliver Davis ◽

...

Keyword(s):

Machine Learning ◽

Causal Inference ◽

Mendelian Randomization ◽

Causal Effect ◽

Association Studies ◽

Genome Wide Association Studies ◽

False Discovery Rates ◽

Level Data ◽

Invalid Instruments ◽

Major Application

AbstractA major application for genome-wide association studies (GWAS) has been the emerging field of causal inference using Mendelian randomization (MR), where the causal effect between a pair of traits can be estimated using only summary level data. MR depends on SNPs exhibiting vertical pleiotropy, where the SNP influences an outcome phenotype only through an exposure phenotype. Issues arise when this assumption is violated due to SNPs exhibiting horizontal pleiotropy. We demonstrate that across a range of pleiotropy models, instrument selection will be increasingly liable to selecting invalid instruments as GWAS sample sizes continue to grow. Methods have been developed in an attempt to protect MR from different patterns of horizontal pleiotropy, and here we have designed a mixture-of-experts machine learning framework (MR-MoE 1.0) that predicts the most appropriate model to use for any specific causal analysis, improving on both power and false discovery rates. Using the approach, we systematically estimated the causal effects amongst 2407 phenotypes. Almost 90% of causal estimates indicated some level of horizontal pleiotropy. The causal estimates are organised into a publicly available graph database (http://eve.mrbase.org), and we use it here to highlight the numerous challenges that remain in automated causal inference.

Download Full-text

Investigating Casual Associations among Gut Microbiota, Metabolites and Neurodegenerative Diseases: A Mendelian Randomization Study

10.21203/rs.3.rs-885681/v1 ◽

2021 ◽

Author(s):

Jin-Tai Yu ◽

Jing Ning ◽

Shu-Yi Huang ◽

Shi-Dong Chen ◽

Yu-Xiang Yang ◽

...

Keyword(s):

Gut Microbiota ◽

Neurodegenerative Diseases ◽

Protective Factor ◽

Mendelian Randomization ◽

Association Studies ◽

Genome Wide Association Studies ◽

Causal Relationships ◽

Genome Wide ◽

Underlying Mechanisms ◽

Lateral Sclerosis

Abstract Background Recent studies had explored that the gut microbiota was associated with neurodegenerative diseases (including Alzheimer’s disease (AD), Parkinson’s disease (PD) and amyotrophic lateral sclerosis (ALS)) through the gut-brain axis, among which metabolic pathways played an important role. However, the underlying causality remained unclear. Our study aimed to evaluate potential causal relationships between gut microbiota, metabolites and neurodegenerative diseases through Mendelian randomization (MR) approach. Methods We selected genetic variants associated with gut microbiota traits (N = 18340) and gut microbiota-derived metabolites (N = 7824) from genome-wide association studies (GWASs). Summary statistics of neurodegenerative diseases were obtained from IGAP (AD: 17008 cases; 37154 controls), IPDGC (PD: 37 688 cases; 141779 controls) and IALSC (ALS: 20806 cases; 59804 controls) respectively. Results A total of 19 gut microbiota traits were found to be causally associated with risk of neurodegenerative diseases, including 1 phylum, 2 classes, 2 orders, 2 families and 12 genera. We found genetically predicted greater abundance of Ruminococcus, at genus level (OR:1.245, 95%CI:1.103,1.405; P = 0.0004) was significantly related to higher risk of ALS. We also found suggestive association between 12 gut microbiome-dependent metabolites and neurodegenerative diseases. For serotonin pathway, our results revealed serotonin as protective factor of PD, and kynurenine as risk factor of ALS. Besides, reduction of glutamine was found causally associated with occurrence of AD. Conclusions Our study firstly applied a two-sample MR approach to detect causal relationships among gut microbiota, gut metabolites and the risk of AD, PD and ALS, and we revealed several causal relationships. These findings may provide new targets for treatment of these neurodegenerative diseases, and may offer valuable insights for further researches on the underlying mechanisms.

Download Full-text

Evaluation of the causal effects of blood lipid levels on gout with summary level GWAS data: two-sample Mendelian randomization and mediation analysis

10.1101/19006296 ◽

2019 ◽

Author(s):

Xinghao Yu ◽

Haimiao Chen ◽

Shuiping Huang ◽

Ping Zeng

Keyword(s):

Mediation Analysis ◽

Large Scale ◽

Mendelian Randomization ◽

Causal Effect ◽

Association Studies ◽

Serum Urate ◽

Mediation Effect ◽

Independent Effect ◽

List Type ◽

Genome Wide Association Studies

AbstractObjectiveMany observational studies have identified that gout patients are often comorbid with dyslipidemia, which is typically characterized by a decrease in high-density lipoprotein cholesterol (HDL) and an increase in triglycerides (TG). However, the relationship between dyslipidemia and gout is still unclear.MethodsWe first performed a two-sample Mendelian randomization (MR) to evaluate the causal effect of four lipid traits on gout and serum urate based on summary association statistics available from large scale genome-wide association studies (up to ∼100,000 for lipid, 69,374 for gout and 110,347 for serum urate). We adopted multivariable Mendelian randomization to estimate the causal effect independently. We also assessed the mediated effect by serum urate between lipids and gout with a mediation analysis. The MR results were validated with extensive sensitive analyses.ResultsGenetically lower HDL was positively associated with the risk of gout and serum urate concentration. Each standard deviation (SD) (∼12.26 mg/dL) increase was genetically associated with an odds ratio of gout of 0.75 (95% CI 0.62 ∼ 0.91, p = 3.31E-3) and with a 0.09 mg/dL (95% CI: -0.12 ∼ -0.05, p = 7.00E-04) decrease in serum urate concentration. Genetically higher TG was positively associated with the serum urate concentration. Each SD (∼112.33 mg/dL) increase was genetically associated with a 0.10 mg/dL (95% CI: 0.06 ∼ 0.14, p = 9.87E-05) increase in serum urate concentration. Those results were robust against various sensitive analyses. In addition, the multivariable Mendelian randomization confirmed the independent effect of HDL and TG on the gout/serum urate after adjustment for the other lipids. Finally, the mediation analysis showed that both HDL and TG could indirectly affect gout morbidity via the pathway of serum urate. The mediation effect accounted for about 13.0% or 28.0% of the total effect of HDL and TG, respectively.ConclusionOur study confirmed the causal associations between HDL/TG and gout/serum urate. Furthermore, the effect of HDL or TG on gout could also be mediated by serum urate.Key MessagesEpidemiological studies have identified an accompanying association between lipid and gout. However, whether the association is causal is unclear.Mendelian randomization with genetic variants as instrumental variables is a useful tool facilitate the validation of a causal relationship for modifiable risk factors.The direct and indirect effects of lipids on gout, controlling for the serum urate concentration, can be estimated by a mediation analysis with serum urate serving as a mediator.We confirmed that elevated HDL levels can directly and indirectly lead to the decreased risk of gout, whereas elevation of TG levels can directly and indirectly elevate the risk of gout.

Download Full-text

Mendelian randomization accounting for correlated and uncorrelated pleiotropic effects using genome-wide summary statistics

10.1101/682237 ◽

2019 ◽

Cited By ~ 5

Author(s):

Jean Morrison ◽

Nicholas Knoblauch ◽

Joseph Marcus ◽

Matthew Stephens ◽

Xin He

Keyword(s):

Genetic Variant ◽

Mendelian Randomization ◽

Association Studies ◽

False Positives ◽

Pleiotropic Effects ◽

Alternative Methods ◽

Genome Wide Association Studies ◽

Causal Relationships ◽

Genome Wide ◽

Summary Effect

AbstractMendelian randomization (MR) is a valuable tool for detecting evidence of causal relationships using genetic variant associations. Opportunities to apply MR are growing rapidly with the number of genome-wide association studies (GWAS) with publicly available results. However, existing MR methods rely on strong assumptions that are often violated, leading to false positives. Many methods have been proposed loosening these assumptions. However, it has remained challenging to account for correlated pleiotropy, which arises when variants affect both traits through a heritable shared factor. We propose a new MR method, Causal Analysis Using Summary Effect Estimates (CAUSE), that accounts for correlated and uncorrelated horizontal pleiotropic effects. We demonstrate in simulations that CAUSE is more robust to correlated pleiotropy than other methods. Applied to traits studied in recent GWAS, we find that CAUSE detects causal relationships with strong literature support and avoids identifying most unlikely relationships. Our results suggest that many pairs of traits identified as causal using alternative methods may be false positives due to horizontal pleiotropy.

Download Full-text