Biology subjects

Hemani, G.

Publications and source records attributed to Hemani, G..

At least 19 recordsLinked to original sources

bioRxiv2026

Identification of Dynamic Genetic Influences on DNA Methylation from Birth to Adulthood

Whether and how genetic regulation of DNA methylation (DNAm) change across the lifespan remains unclear. Here, we map age-dependent methylation quantitative trait loci (longitudinal mQTLs) using linear mixed models applied to repeated blood DNAm measures from birth, childhood, adolescence and adulthood in the Avon Longitudinal Study of Parents and Children. We identify 2,210 longitudinal mQTLs (2,393 SNP-CpG pairs; 7.3% trans) and observe consistent genotype-by-age effects in two independent cohorts of diverse ancestries (Pearsons r = 0.85 in the Generation R Study; r = 0.56 in the Drakenstein Child Health Study). Longitudinal mQTLs show increasing effects with age at half of loci and associations with multiple phenotypes. CpGs with longitudinal mQTLs are more heritable and enriched in regulatory elements and pathways related to multicellular organism development and cell adhesion. These results chart dynamic genetic influences on the human methylome and provide a novel perspective on epigenetic regulation.

genomics↗

bioRxiv2025

Simulated sample splitting approach to address biases due to instrument selection and participant overlap in two-sample Mendelian Randomization studies

Mendelian randomization (MR) is a popular statistical technique that uses genetic variants to explore causal relationships in observational epidemiology. Summary-level MR, the most common form, relies on published GWAS summary statistics to estimate causal effects between exposures and outcomes. However, empirical analyses tend to ignore issues relating to Winners Curse of instrument effects, weak instrument bias and sample overlap. Our simulations and empirical examinations using the UK Biobank indicate that such mechanisms can induce substantial bias in routine MR approaches. We propose MR Simulated Sample Splitting (MR-SimSS), a novel method that corrects this bias requiring no additional data beyond the exposure and outcome GWAS summary statistics under examination. It operates by simulating statistically independent sets of summary statistics, analogous to what would be produced by splitting the individual-level data into independent subsets, which can then be plugged into existing pleiotropy-robust MR methods. With sufficient instrument variants, MR-SimSS is robust to a range of sample overlap scenarios, providing a practical and modular solution to Winners Curse and weak instrument bias. Author summaryA central challenge in epidemiology is determining whether an observed association reflects a true cause- and-effect relationship. Mendelian randomization (MR) addresses this by using genetic variants as natural experiments to test whether a particular trait or exposure genuinely influences disease risk. However, when the same genetic data are used both to select and to estimate genetic instruments, MR results can become biased due to a phenomenon known as the Winners Curse. This problem, along with weak instruments and sample overlap between datasets, can distort causal estimates even in large studies. We introduce MR Simulated Sample Splitting (MR-SimSS), a new framework that overcomes these issues using only publicly available genome-wide association study (GWAS) summary statistics. MR-SimSS works by statistically simulating independent subsets of the data, without requiring access to individual-level information, allowing existing MR methods to be applied without bias. Through extensive simulations and analyses using UK Biobank data, we show that MR-SimSS provides more accurate and reliable causal estimates, offering a practical tool for robust causal inference in modern genetic epidemiology.

genetics↗

bioRxiv2025

Robust inference and widespread genetic correlates from a large-scale genetic association study of human personality

Personality traits describe stable differences in how individuals think, feel, and behave and how they interact with and experience their social and physical environments. We assemble data from 46 cohorts including 611K-1.14M participants with European-like and African-like genomes for genome-wide association studies (GWAS) of the Big Five personality traits (extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience), and data from 51K participants for within-family GWAS. We identify 1,257 lead genetic variants associated with personality, including 823 novel variants. Common genetic variants explain 4.8%-9.3% of the variance in each trait, and 10.5%-16.2% accounting for measurement unreliability. Genetic effects on personality are highly consistent across geography, reporter (self vs. close other), age group, and measurement instrument, and we find minimal spousal assortment for personality in recent history. In stark contrast to many other social and behavioral traits, within-family GWAS and polygenic index analyses indicate little to no shared environmental confounding in genetic associations with personality. Polygenic prediction, genetic correlation, and Mendelian randomization analyses indicate that personality genetics have widespread, potentially causal associations with a wide range of consequential behaviors and life outcomes. The genetic architecture of personality is robust and fundamental to being a human.

genomics↗

bioRxiv2022

Review and further developments in statistical corrections for Winner's Curse in genetic association studies

Genome-wide association studies (GWAS) are commonly used to identify genomic variants that are associated with complex traits, and estimate the magnitude of this association for each variant. However, it has been widely observed that the association estimates of variants tend to be lower in a replication study than in the study that discovered those associations. A phenomenon known as Winners Curse is responsible for this upward bias present in association estimates of significant variants in the discovery study. We review existing Winners Curse correction methods which require only GWAS summary statistics in order to make adjustments. In addition, we propose modifications to improve existing methods and propose a novel approach which uses the parametric bootstrap. We evaluate and compare methods, first using a wide variety of simulated data sets and then, using real data sets for three different traits. The metric, estimated mean squared error (MSE) over significant SNPs, was primarily used for method assessment. Our results indicate that widely used conditional likelihood based methods tend to perform poorly. The other considered methods behave much more similarly, with our proposed bootstrap method demonstrating very competitive performance. To complement this review, we have developed an R package, winnerscurse which can be used to implement these various Winners Curse adjustment methods to GWAS summary statistics. Author SummaryA genome-wide association study is designed to analyse many common genetic variants in thousands of samples and identify which variants are associated with a trait of interest. It provides estimates of association strength for each variant and variants are classified as associated if their test statistics obtained in the study pass a chosen significance threshold. However, due to a phenomenon known as Winners Curse, the association estimates of these significant variants tend to be upward biased and greater in magnitude than their true values. Naturally, this bias has adverse consequences for downstream statistical techniques which use these estimates. In this paper, we look at current methods which have been designed to combat Winners Curse and propose modifications to these methods in order to improve performance. Using a wide variety of simulated data sets as well as real data, we perform a thorough evaluation of these methods. We use a metric which allows us to identify which methods, on average, produce adjusted estimates for significant variants that are closest to the true values. To accompany our work, we have created an R package, winnerscurse, which allows users to easily apply Winners Curse correction methods to their data sets.

genetics↗

bioRxiv2022

An efficient and robust tool for colocalisation: Pair-wise Conditional and Colocalisation (PWCoCo)

Genetic colocalisation is an important tool to test for shared genetic aetiology and is commonly used to strengthen causal inference in genetic studies of molecular traits and drug targets. However, the single causal variant assumption of the original colocalization method is a considerable limitation in genomic regions with multiple causal effects. We integrated conditional analyses (GCTA-COJO) and colocalisation analyses (coloc), into a novel analysis tool called Pair-Wise Conditional Colocalization (PWCoCo). PWCoCo performs conditional analyses to identify independent signals for the two tested traits in a genomic region and then conducts colocalisation of each pair of conditionally independent signals for the two traits using summary-level data. This allows for the stringent single-variant assumption to hold for each pair of colocalisation analysis. We found that the computational efficiency of PWCoCo is on average better than colocalisation with Sum of Single Effects Regression using Summary Stats (SuSiE-RSS), with greater gains in efficiency for high-throughput analysis. In a case study using GWAS data for multiple sclerosis and brain cortex-derived eQTLs (MetaBrain), we recapitulated all previously identified genes, which showcased the robustness of the method. We further found colocalisation evidence for secondary signals in nine additional loci, which was not identifiable in conventional GWAS and/or colocalisation. PWCoCo offers key improvements over existing methods, including: (1) robust colocalisation when the single variant assumption is violated; (2) independent colocalisation of secondary signals, which enables identification of novel disease-causing variants; (3) an easy-to-use and computationally efficient tool to test for colocalisation of high-dimensional omics data.

bioinformatics↗

bioRxiv2022

The Norwegian Mother, Father, and Child cohort study (MoBa) genotyping data resource: MoBaPsychGen pipeline v.1

BackgroundThe Norwegian Mother, Father, and Child Cohort Study (MoBa) is a population-based pregnancy cohort, which includes approximately 114,500 children, 95,200 mothers, and 75,200 fathers. Genotyping of MoBa has been conducted through multiple research projects, spanning several years; using varying selection criteria, genotyping arrays, and genotyping centres. MoBa contains numerous interrelated families, which necessitated the implementation of a family-based quality control (QC) pipeline that verifies and accounts for diverse types of relatedness. MethodsThe MoBaPsychGen pipeline, comprising pre-imputation QC, phasing, imputation, and post-imputation QC, was developed based on current best-practice protocols and implemented to account for the complex structure of the MoBa genotype data. The pipeline includes QC on both single nucleotide polymorphism (SNP) and individual level. Phasing and imputation were performed using the publicly available Haplotype Reference Consortium release 1.1 panel as a reference. Information from the Medical Birth Registry of Norway and MoBa questionnaires were used to identify biological sex, year of birth, reported parent-offspring (PO) relationships, and multiple births (only available in the offspring generation). ResultsIn total, 207,569 unique individuals (90% of the unique individuals included in the study) and 6,981,748 autosomal SNPs passed the MoBaPsychGen pipeline. A further 174,462 chromosome X and 3,200 PAR SNPs are available in a subset of these individuals (N = 204,913 and 135,593, respectively). The relatedness checks performed throughout the pipeline allowed identification of within-generation and across-generation first-degree, second-degree, and third-degree relatives. The individuals passing post-imputation QC comprised 64,471 families ranging in size from singletons to 84 unique individuals (singletons are included as families as other family members may not have been genotyped, imputed, or passed post-imputation QC). The relationships identified include 287 monozygotic twin pairs, 22,884 full siblings, 117,004 PO pairs, 23,299 second-degree relative pairs, and 10,828 third-degree relative pairs. DiscussionMoBa contains a highly complex relatedness structure, with a variety of family structures including singletons, PO duos, full (mother, father, child) PO trios, nuclear families, blended families, and extended families. The availability of robustly quality-controlled genetic data for such a large cohort with a unique extended family structure will allow many novel research questions to be addressed. Furthermore, the MoBaPsychGen pipeline has potential utility in similar cohorts.

genetics↗

bioRxiv2022

DNA co-methylation has a stable structure and is related to specific aspects of genome regulation

DNA methylation (DNAm) is influenced by genetic and environmental factors, and can be used to understand interindividual variability in genomic regulation. Co-methylation between DNAm sites is a known phenomenon, but the architecture of relationships between the approximately 450,000 (450k) sites commonly measured in epidemiological studies has not been described. We investigate whether interindividual co-methylation structure amongst the 450k sites changes with age, whether it differs between UK-born White (n=849, 910, 921 and 424) and Pakistani ancestry (n=439) individuals, and how it relates to genome regulation. We find stability between birth and adolescence, across cohorts, and between two ethnic groups. Highly correlated DNAm sites in close proximity are heritable, but these relationships are weakly influenced by nearby genetic variants, and are enriched for transcription factor (TF) binding sites related to regulation of short RNAs transcribed by RNA polymerase III. Highly correlated sites that are distant, or on different chromosomes (in trans), are driven by common and unique environmental factors, with methylation at these sites less likely to be driven by genotype. Trans co-methylated DNAm sites are enriched for multiple TF binding sites and for inter-chromosomal chromatin contact sites, suggesting DNA co-methylation of distant sites may relate to long-range cooperative TF interactions. We conclude that DNA co-methylation has a stable structure from birth to adolescence, and between UK-born White and Pakistani individuals. This stable structure might have implications for future design and interpretation of epigenetic studies. We hypothesise that co-methylation may have roles in genome regulation in humans, including 3D chromatin architecture.

bioinformatics↗

bioRxiv2022

Evaluating the potential benefits and pitfalls of combining protein and expression quantitative trait loci in evidencing drug targets

Molecular quantitative trait loci (molQTL), which can provide functional evidence on the mechanisms underlying phenotype-genotype associations, are increasingly used in drug target validation and safety assessment. In particular, protein abundance QTLs (pQTLs) and gene expression QTLs (eQTLs) are the most commonly used for this purpose. However, questions remain on how to best consolidate results from pQTLs and eQTLs for target validation. In this study, we combined blood cell-derived eQTLs and plasma-derived pQTLs to form QTL pairs representing each gene and its product. We performed a series of enrichment analyses to identify features of QTL pairs that provide consistent evidence for drug targets based on the concordance of the direction of effect of the pQTL and eQTL. We repeated these analyses using eQTLs derived in 49 tissues. We found that 25-30% of blood-cell derived QTL pairs have discordant effects. The difference in tissues of origin for molecular markers contributes to, but is not likely a major source of, this observed discordance. Finally, druggable genes were as likely to have discordant QTL pairs as concordant. Our analyses suggest combining and consolidating evidence from pQTLs and eQTLs for drug target validation is crucial and should be done whenever possible, as many potential drug targets show discordance between the two molecular phenotypes that could be misleading if only one is considered. We also encourage investigating QTL tissue-specificity in target validation applications to help identify reasons for discordance and emphasise that concordance and discordance of QTL pairs across tissues are both informative in target validation.

genetics↗

bioRxiv2021

Evidence of positive and negative selection associated with DNA methylation

Signatures of negative selection are pervasive amongst complex traits and diseases. However, it is unclear whether such signatures exist for DNA methylation (DNAm) that has been proposed to have a functional role in disease. We estimate polygenicity, SNP-based heritability and model the joint distribution of effect size and minor allele frequency (MAF) to estimate a selection coefficient (S) for 2000 heritable DNAm sites in 1774 individuals from the Avon Longitudinal Study of Parents and Children. Additionally, we estimate S for meta stable epi alleles and DNAm sites associated with aging and mortality, birthweight and body mass index. Quantification of MAF-dependent genetic architectures estimated from genotype and DNAm reveal evidence of positive (S > 0) and negative selection (S < 0) and confirm previous evidence of negative selection for birthweight. Evidence of both negative and positive selection highlights the role of DNAm as an intermediary in multiple biological pathways with competing function.

genetics↗

bioRxiv2020

Exploring the variance in complex traits captured by DNA methylation assays

Following years of epigenome-wide association studies (EWAS), traits analysed to date tend to yield few associations. Reinforcing this observation, we conducted EWAS on 400 traits and 16 yielded at least one association at the conventional significance threshold (P<1x10-7). To investigate why EWAS yield is low, we formally estimated the proportion of phenotypic variation captured by 421,693 blood derived DNA methylation markers (h2EWAS) across all 400 traits. The mean h2EWAS was zero, with evidence for regular cigarette smoking exhibiting the largest association with all markers (h2EWAS=0.42) and the only one surpassing a false discovery rate < 0.1. Though underpowered to determine the h2EWAS value for any one trait, h2EWAS was predictive of the number of EWAS hits across the traits analysed (AUC=0.7). Modelling the contributions of the methylome on a per-site versus a per-region basis gave varied h2EWAS estimates (r=0.47) but neither approach obtained substantially higher model fits across all traits. Our analysis indicates that most complex traits do not heavily associate with markers commonly measured in EWAS within blood. However, it is likely DNA methylation does capture variation in some traits and h2EWAS may be a reasonable way to prioritise traits that are likely to yield associations.

genetics↗

bioRxiv2020

The MRC IEU OpenGWAS data infrastructure

Data generated by genome-wide association studies (GWAS) are growing fast with the linkage of biobank samples to health records, and expanding capture of high-dimensional molecular phenotypes. However the utility of these efforts can only be fully realised if their complete results are collected from their heterogeneous sources and formats, harmonised and made programmatically accessible. Here we present the OpenGWAS database, an open source, open access, scalable and high-performance cloud-based data infrastructure that imports and publishes complete GWAS summary datasets and metadata for the scientific community. Our import pipeline harmonises these datasets against dbSNP and the human genome reference sequence, generates summary reports and standardises the format of results and metadata. Users can access the data via a website, an application programming interface, R and Python packages, and also as downloadable files that can be rapidly queried in high performance computing environments. OpenGWAS currently contains 126 billion genetic associations from 14,582 complete GWAS datasets representing a range of different human phenotypes and disease outcomes across different populations. We developed R and Python packages to serve as conduits between these GWAS data sources and a range of available analytical tools, enabling Mendelian randomization, genetic colocalisation analysis, fine mapping, genetic correlation and locus visualisation. OpenGWAS is freely accessible at https://gwas.mrcieu.ac.uk, and has been designed to facilitate integration with third party analytical tools.

genetics↗

bioRxiv2020

EpiGraphDB: a database and data mining platform for health data science

MotivationThe wealth of data resources on human phenotypes, risk factors, molecular traits and therapeutic interventions presents new opportunities for population health sciences. These opportunities are paralleled by a growing need for data integration, curation and mining to increase research efficiency, reduce mis-inference and ensure reproducible research. ResultsWe developed EpiGraphDB (https://epigraphdb.org/), a graph database containing an array of different biomedical and epidemiological relationships and an analytical platform to support their use in human population health data science. In addition, we present three case studies that illustrate the value of this platform. The first uses EpiGraphDB to evaluate potential pleiotropic relationships, addressing mis-inference in systematic causal analysis. In the second case study we illustrate how protein-protein interaction data offer opportunities to identify new drug targets. The final case study integrates causal inference using Mendelian randomization with relationships mined from the biomedical literature to "triangulate" evidence from different sources. AvailabilityThe EpiGraphDB platform is openly available at https://epigraphdb.org. Code for replicating case study results is available at https://github.com/MRCIEU/epigraphdb as Jupyter notebooks using the API, and https://mrcieu.github.io/epigraphdb-r using the R package. Contactyi6240.liu@bristol.ac.uk, ben.elsworth@bristol.ac.uk, Tom.Gaunt@bristol.ac.uk

bioinformatics↗

bioRxiv2020

The use of negative control outcomes in Mendelian Randomisation to detect potential population stratification or selection bias.

A key assumption of Mendelian randomisation (MR) analysis is that there is no association between the genetic variants used as instruments and the outcome other than through the exposure of interest. Two ways in which this assumption can be violated are through population stratification and selection bias which can introduce confounding of the relationship between the genetic variants and the outcome and so induce an association between them. Negative control outcomes are increasingly used to detect unobserved confounding in observational epidemiological studies. Here we consider the use of negative control outcomes in MR studies. As a negative control outcome in an MR study we propose the use of phenotypes which are determined before the exposure and outcome but which are likely to be subject to the same confounding as the exposure or outcome of interest. We illustrate our method with a two-sample MR analysis of a preselected set of exposures on self-reported tanning ability and hair colour. Our results show that, of the 33 exposures considered, GWAS studies of adiposity and education related traits are likely to be subject to population stratification and/or selection bias that is not controlled for through adjustment and so any MR study including these traits may be subject to bias that cannot be identified through standard pleiotropy robust methods.

genetics↗

bioRxiv2020

The variant call format provides efficient and robust storage of GWAS summary statistics

Genome-wide association study (GWAS) summary statistics are a fundamental resource for a variety of research applications 1-6. Yet despite their widespread utility, no common storage format has been widely adopted, hindering tool development and data sharing, analysis and integration. Existing tabular formats 7,8 often ambiguously or incompletely store information about genetic variants and their associations, and also lack essential metadata increasing the possibility of errors in data interpretation and post-GWAS analyses. Additionally, data in these formats are typically not indexed, requiring the whole file to be read which is computationally inefficient. To address these issues, we propose an adaptation of the variant call format9 (GWAS-VCF) and have produced a suite of open-source tools for using this format in downstream analyses. Simulation studies determine GWAS-VCF is 9-46x faster than tabular alternatives when extracting variant(s) by genomic position. Our results demonstrate the GWAS-VCF provides a robust and performant solution for sharing, analysis and integration of GWAS data. We provide open access to over 10,000 complete GWAS summary datasets converted to this format (available from: https://gwas.mrcieu.ac.uk).

genetics↗

bioRxiv2020

Causal Inference for Heritable Phenotypic Risk Factors Using Heterogeneous Genetic Instruments

Over a decade of genome-wide association studies (GWAS) have led to the finding of extreme polygenicity of complex traits. The phenomenon that "all genes affect every complex trait" complicates Mendelian Randomization (MR) studies, where natural genetic variations are used as instruments to infer the causal effect of heritable risk factors. We reexamine the assumptions of existing MR methods and show how they need to be clarified to allow for pervasive horizontal pleiotropy and heterogeneous effect sizes. We propose a comprehensive framework GRAPPLE to analyze the causal effect of target risk factors with heterogeneous genetic instruments and identify possible pleiotropic patterns from data. By using GWAS summary statistics, GRAPPLE can efficiently use both strong and weak genetic instruments, detect the existence of multiple pleiotropic pathways, determine the causal direction and perform multivariable MR to adjust for confounding risk factors. With GRAPPLE, we analyze the effect of blood lipids, body mass index, and systolic blood pressure on 25 disease outcomes, gaining new information on their causal relationships and the potential pleiotropic pathways.

genetics↗

bioRxiv2020

Investigating DNA methylation as a potential mediator between pigmentation genes, pigmentary traits and skin cancer

BackgroundIncidence rates for melanoma and non-melanoma skin cancer (NMSC), which includes basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), have been steadily increasing in all populations. Populations of European ancestry exhibit the highest rates and therefore, have been widely studied. Pigmentation characteristics are well-known risk factors for skin cancer, particularly fair skin, red hair, blue eyes and the inability to tan. Polymorphisms in established pigmentation-related genes have been associated with these traits and with an increased risk of malignancy. However, the functional relationship between genetic variation and disease is still unclear, with the exception of red hair colour variants in the melanocortin 1 receptor (MC1R) gene. ObjectivesThe aim of this study was to explore the possibility that non-coding pigmentation SNPs are associated with pigmentary traits and skin cancer via DNA methylation (DNAm). Methods and ResultsUsing a meta-GWAS of whole blood DNAm from 36 European cohorts (N=27,750; the Genetics of DNA Methylation Consortium, GoDMC), we found that 19 out of 27 pigmentation-associated SNPs distributed within 10 genes (ASIP, BNC2, IRF4, HERC2, MC1R, OCA2, SLC24A4, SLC24A5, SLC45A2, TYR) were associated with 391 DNAm sites across 30 genomic regions. We selected 25 DNAm sites for further analysis. We examined the effect of the chosen DNAm sites on pigmentation traits, sun exposure phenotypes, and skin cancer, and on gene expression in whole blood. We found an association of decreased DNAm at cg07402062 with red hair in the Avon Longitudinal Study of Parents and Children (ALSPAC), and a strong positive association of DNAm at this and correlated sites with higher expression of SPIRE2. Additionally, we investigated the association of gene expression in skin with pigmentation traits and skin cancer. The expression of ASIP, FAM83C, NCOA6, CDK10, and EXOC2 was associated with hair colour, whilst that of ASIP and CDK10 also had an effect on melanoma and BCC. ConclusionsOur results indicate that DNAm and expression of genes in the 16q24.3 and 20q11.22 regions, deserve to be further investigated as potential mediators of the relationship between genetic variants, pigmentation/sun exposure phenotypes, and some types of skin cancer.

genetics↗

bioRxiv2020

Assortative mating and within-spouse pair comparisons

Comparisons between cohabitating spouses have been proposed as an aetiological design method to reduce confounding and evaluate effects of the shared adulthood environment. However, assortative mating, a phenomenon where individuals select phenotypically similar mates, could distort associations. We evaluated the use of spousal comparisons, as in the within-spouse pair (WSP) model, for aetiological epidemiological research. Using directed acyclic graphs and simulations, we demonstrated that the WSP model can reduce confounding if spouses are correlated for an unmeasured confounder, but that WSP comparisons are susceptible to collider bias induced by assortative mating. Empirical analyses using spouse pairs in UK Biobank found evidence that genetic association estimates from the WSP model are attenuated compared to random pairs for single nucleotide polymorphisms (SNPs) associated with height (shrinkage: 23%; 95% CI 20%, 25%), educational attainment (74%; 95% CI 66%, 81%) and body mass index (23%; 95% CI 14%, 32%) as well as for an alcohol consumption SNP (29%, 95% CI 5%, 46%). Some of these attenuations are likely to reflect effects of assortative mating because height and educational attainment are unlikely to be strongly influenced by the adulthood environment. In contrast, effect estimates of increasing age on coronary artery disease and systolic blood pressure were found to be concordant between random and spouse pairs. Assortative mating is likely to induce phenotypic and genetic structure between an individual and their spouse which complicates the interpretation of spousal comparisons in an aetiological context. A further consideration is that the joint participation of non-independent spouses in cohort studies could induce selection bias.

genetics↗

bioRxiv2019

The association of DNA methylation with body mass index: distinguishing between predictors and biomarkers

BackgroundDNA methylation is associated with body mass index (BMI), but it is not clear if methylation scores are biomarkers for extant BMI, or predictive of future BMI. Here we explore the causal nature and predictive utility of DNA methylation measured in peripheral blood with BMI and cardiometabolic traits. MethodsAnalyses were conducted across the life course using the ARIES cohort of mothers (n=792) and children (n=906), for whom DNA methylation and genetic profiles and BMI at multiple time points (3 in children at birth, in childhood and in adolescence, 2 in mothers during pregnancy and in middle age) were available. Genetic and DNA methylation scores for BMI were derived using published associations between BMI and DNA methylation and genotype. Causal relationships between methylation and BMI were assessed using Mendelian randomisation and cross-lagged models. ResultsThe DNA methylation scores in adult women explained 10% of extant BMI variance. However, less extant variance was explained by scores generated in the same women during pregnancy (2% BMI variance) and in older children (15-17 years; 3% BMI variance). Similarly, little extant variance was explained in younger children (at birth and at 7 years; 1% and 2%, respectively). These associations remained following adjustment for smoking exposure and education levels. The DNA methylation score was found to be a poor predictor of future BMI using linear and cross-lagged models, suggesting that DNA methylation variation does not cause later variation in BMI. However, there was some evidence to suggest that BMI is predictive of later DNA methylation. Mendelian randomisation analyses also support this direction of effect, although evidence is weak. Finally, we find that DNA methylation scores for BMI are associated with extant cardiometabolic traits independently of BMI and genetic score. ConclusionThe age-specific nature of DNA methylation associations with BMI, lack of causal relationship, and limited predictive ability of future BMI, indicate that DNA methylation is likely influenced by BMI and might more accurately be considered a biomarker of BMI and related outcomes than a predictor. Future epigenome-wide association studies may benefit from further examining associations between early DNA methylation and later health outcomes.

genetics↗