Batch effects and pathway analysis: two potential perils in cancer studies involving DNA methylation array analysis

Does biological age mediate the relationship between childhood adversity and depression? Insights from the Detroit Neighborhood Health Study

The link between childhood adversity and adulthood depression is well-established; however, the underlying mechanisms are still being explored. Recent research suggests biological age may mediate the relationship between childhood adversity and depression in later life. This study examines if biological age mediates the relationship between childhood adversity and depression symptoms using an expanded set of biological age measures in an urban population-based cohort. Data from waves 1-3 of the Detroit Neighborhood Health Study (DNHS) were used in this analysis. Questions about abuse during childhood were coded to form a childhood adversity score similar to the Adverse Childhood Experience measure. Multiple dimensions of biological age, defined as latent variables, were considered, including systemic biological age (GrimAge, PhenoAge), epigenetic age (Horvath, SkinBlood), and immune age (cytomegalovirus, herpes simplex virus type 1, C-reactive protein, interleukin-6). Depression symptoms, modeled as a latent variable, were captured through the Patient Health Questionnaire-9 (PHQ-9). Models were adjusted for age, gender, race, parent education, and past depressive symptoms. Total and direct effects of childhood adversity on depression symptoms and indirect effects mediated by biological age were estimated. For total and direct effects, we observed a dose-dependent relationship between cumulative childhood adversity and depression symptoms, with emotional abuse being particularly influential. However, contrary to prior studies, in this sample, we found few direct effects of childhood adversity on biological age or biological age on depression symptoms and no evidence of mediation through the measures of biological age considered in this study. Further research is needed to understand how childhood maltreatment experiences are embodied to influence health and wellness.

Sex-specific DNA methylation in saliva from the multi-ethnic Future of Families and Child Wellbeing Study

The prevalence and severity of many diseases differs by sex, potentially due to sex-specific patterns in DNA methylation. Autosomal sex-specific differences in DNA methylation have been observed in cord blood and placental tissue but are not well studied in saliva or in diverse populations. We sought to characterize sex-specific DNA methylation on autosomal chromosomes in saliva samples from children in the Future of Families and Child Wellbeing Study, a multi-ethnic prospective birth cohort containing an oversampling of Black, Hispanic and low-income families. DNA methylation from saliva samples was analysed on 796 children (50.6% male) at both ages 9 and 15 with DNA methylation measured using the Illumina HumanMethylation 450k array. An epigenome-wide association analysis of the age 9 samples identified 8,430 sex-differentiated autosomal DNA methylation sites (P < 2.4 × 10-7), of which 76.2% had higher DNA methylation in female children. The strongest sex-difference was in the cg26921482 probe, in the AMDHD2 gene, with 30.6% higher DNA methylation in female compared to male children (P < 1 × 10-300). Treating the age 15 samples as an internal replication set, we observed highly consistent results between the ages 9 and 15 measurements, indicating stable and replicable sex-differentiation. Further, we directly compared our results to previously published DNA methylation sex differences in both cord blood and saliva and again found strong consistency. Our findings support widespread and robust sex-differential DNA methylation across age, human tissues, and populations. These findings help inform our understanding of potential biological processes contributing to sex differences in human physiology and disease.

Early-life exposure to cigarette smoke primes lung function and DNA methylation changes at Cyp1a1 upon exposure later in life

Prenatal and early-life exposure to cigarette smoke (CS) has repeatedly been shown to induce stable, long-term changes in DNA methylation (DNAm) in offspring. It has been hypothesized that these changes might be functionally related to the known outcomes of prenatal and early-life CS exposure, which include impaired lung development, altered lung function, and increased risk of asthma and wheeze. However, to date, few studies have examined DNAm changes induced by prenatal CS in tissues of the lung, and even fewer have attempted to examine the specific influences of prenatal versus early postnatal exposures. Here, we have established a mouse model of CS exposure which isolates the effects of prenatal and early postnatal CS exposures in early life. We have used this model to measure the effects of prenatal and/or postnatal CS exposures on lung function and immune cell infiltration as well as DNAm and expression of Cyp1a1, a candidate gene previously observed to demonstrate DNAm differences on CS exposure in humans. Our study revealed that exposure to CS prenatally and in the early postnatal period causes long-lasting differences in offspring lung function, gene expression, and lung Cyp1a1 DNAm, which wane over time but are reestablished on reexposure to CS in adulthood. This study creates a testable mouse model that can be used to investigate the effects of prenatal and early postnatal CS exposures and will contribute to the design of intervention strategies to mediate these detrimental effects.NEW & NOTEWORTHY Here, we isolated effects of prenatal from early postnatal cigarette smoke and showed that exposure to cigarette smoke early in life causes changes in offspring DNA methylation at Cyp1a1 that last through early adulthood but not into late adulthood. We also showed that smoking in adulthood reestablished these DNA methylation patterns at Cyp1a1, suggesting that a mechanism other than DNA methylation results in long-term memory associated with early-life cigarette smoke exposures at this gene.

Normalization of Large-Scale Transcriptome Data Using Heuristic Methods

In this study, we introduce an artificial intelligent method for addressing the batch effect of a transcriptome data. The method has several clear advantages in comparison with the alternative methods presently in use. Batch effect refers to the discrepancy in gene expression data series, measured under different conditions. While the data from the same batch (measurements performed under the same conditions) are compatible, combining various batches into 1 data set is problematic because of incompatible measurements. Therefore, it is necessary to perform correction of the combined data (normalization), before performing biological analysis. There are numerous methods attempting to correct data set for batch effect. These methods rely on various assumptions regarding the distribution of the measurements. Forcing the data elements into pre-supposed distribution can severely distort biological signals, thus leading to incorrect results and conclusions. As the discrepancy between the assumptions regarding the data distribution and the actual distribution is wider, the biases introduced by such “correction methods” are greater. We introduce a heuristic method to reduce batch effect. The method does not rely on any assumptions regarding the distribution and the behavior of data elements. Hence, it does not introduce any new biases in the process of correcting the batch effect. It strictly maintains the integrity of measurements within the original batches.

Tissue- and ethnicity-independent hypervariable DNA methylation states show evidence of establishment in the early human embryo

We analysed DNA methylation data from 30 datasets comprising 3474 individuals, 19 tissues and 8 ethnicities at CpGs covered by the Illumina450K array. We identified 4143 hypervariable CpGs ('hvCpGs') with methylation in the top 5% most variable sites across multiple tissues and ethnicities. hvCpG methylation was influenced but not determined by genetic variation, and was not linked to probe reliability, epigenetic drift, age, sex or cell heterogeneity effects. hvCpG methylation tended to covary across tissues derived from different germ-layers and hvCpGs were enriched for proximity to ERV1 and ERVK retrovirus elements. hvCpGs were also enriched for loci previously associated with periconceptional environment, parent-of-origin-specific methylation, and distinctive methylation signatures in monozygotic twins. Together, these properties position hvCpGs as strong candidates for studying how stochastic and/or environmentally influenced DNA methylation states which are established in the early embryo and maintained stably thereafter can influence life-long health and disease.

Arsenic exposure and human blood DNA methylation and hydroxymethylation profiles in two diverse populations from Bangladesh and Spain

BACKGROUND

Associations of arsenic (As) with the sum of 5-mC and 5-hmC levels have been reported; however, As exposure-related differences of the separated 5-mC and 5-hmC markers have rarely been studied.

METHODS

In this study, we evaluated the association of arsenic exposure biomarkers and 5-mC and 5-hmC in 30 healthy men (43-55 years) from the Aragon Workers Health Study (AWHS) (Spain) and 31 healthy men (31-50 years) from the Folic Acid and Creatinine Trial (FACT) (Bangladesh). We conducted 5-mC and 5-hmC profiling using Infinium MethylationEPIC arrays, on paired standard and modified (ox-BS in AWHS and TAB in FACT) bisulfite converted blood DNA samples.

RESULTS

The median for the sum of urine inorganic and methylated As species (ΣAs) (μg/L) was 12.5 for AWHS and 89.6 for FACT. The median of blood As (μg/L) was 8.8 for AWHS and 10.2 for FACT. At a statistical significance p-value cut-off of 0.01, the differentially methylated (DMP) and hydroxymethylated (DHP) positions were mostly located in different genomic sites. Several DMPs and DHPs were consistently found in AWHS and FACT both for urine ΣAs and blood models, being of special interest those attributed to the DIP2C gene. Three DMPs (annotated to CLEC12A) for AWHS and one DHP (annotated to NPLOC4) for FACT remained statistically significant after false discovery rate (FDR) correction. Pathways related to chronic diseases including cardiovascular, cancer and neurological were enriched.

CONCLUSIONS

While we identified common 5-hmC and 5-mC signatures in two populations exposed to varying levels of inorganic As, differences in As-related epigenetic sites across the study populations may additionally reflect low and high As-specific associations. This work contributes a deeper understanding of potential epigenetic dysregulations of As. However, further research is needed to confirm biological consequences associated with DIP2C epigenetic regulation and to investigate the role of 5-hmC and 5-mC separately in As-induced health disorders at different exposure levels.

Evaluation of critical data processing steps for reliable prediction of gene co-expression from large collections of RNA-seq data

Motivation

Gene co-expression analysis is an attractive tool for leveraging enormous amounts of public RNA-seq datasets for the prediction of gene functions and regulatory mechanisms. However, the optimal data processing steps for the accurate prediction of gene co-expression from such large datasets remain unclear. Especially the importance of batch effect correction is understudied.

Results

We processed RNA-seq data of 68 human and 76 mouse cell types and tissues using 50 different workflows into 7,200 genome-wide gene co-expression networks. We then conducted a systematic analysis of the factors that result in high-quality co-expression predictions, focusing on normalization, batch effect correction, and measure of correlation. We confirmed the key importance of high sample counts for high-quality predictions. However, choosing a suitable normalization approach and applying batch effect correction can further improve the quality of co-expression estimates, equivalent to a >80% and >40% increase in samples. In larger datasets, batch effect removal was equivalent to a more than doubling of the sample size. Finally, Pearson correlation appears more suitable than Spearman correlation, except for smaller datasets.

Conclusion

A key point for accurate prediction of gene co-expression is the collection of many samples. However, paying attention to data normalization, batch effects, and the measure of correlation can significantly improve the quality of co-expression estimates.

Omixer: multivariate and reproducible sample randomization to proactively counter batch effects in omics studies

MOTIVATION

Batch effects heavily impact results in omics studies, causing bias and false positive results, but software to control them preemptively is lacking. Sample randomization prior to measurement is vital for minimizing these effects, but current approaches are often ad hoc, poorly documented and ill-equipped to handle multiple batches and outcomes.

RESULTS

We developed Omixer-a Bioconductor package implementing multivariate and reproducible sample randomization for omics studies. It proactively counters correlations between technical factors and biological variables of interest by optimizing sample distribution across batches.

AVAILABILITYAND IMPLEMENTATION

Omixer is available from Bioconductor at http://bioconductor.org/packages/release/bioc/html/Omixer.html. Scripts and data used to generate figures available upon request.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Effect of prenatal exposure to phthalates on epigenome-wide DNA methylations in cord blood and implications for fetal growth: The Hokkaido Study on Environment and Children's Health

Prenatal exposure to phthalates negatively affects the offspring's health. In particular, epigenetic alterations, such as DNA methylation, may connect phthalate exposure with health outcomes. Here, we evaluated the association of di-2-ethylhexyl phthalate (DEHP) exposure in utero with cord blood epigenome-wide DNA methylation in 203 mother-child pairs enrolled in the Hokkaido Study on Environment and Children's Health, using the Illumina HumanMethylation450 BeadChip. Epigenome-wide association analysis demonstrated the predominant positive associations between the levels of the primary metabolite of DEHP, mono(2-ethylhexyl) phthalate (MEHP), in maternal blood and DNA methylation levels in cord blood. The genes annotated to the CpGs positively associated with MEHP levels were enriched for pathways related to metabolism, the endocrine system, and signal transduction. Among them, methylation levels of CpGs involved in metabolism were inversely associated with the offspring's ponderal index (PI). Further, clustering and mediation analyses suggested that multiple increased methylation changes may jointly mediate the association of DEHP exposure in utero with the offspring's PI at birth. Although further studies are required to assess the impact of these changes, this study suggests that differential DNA methylation may link phthalate exposure in utero to fetal growth and further imply that DNA methylation has predictive value for the offspring's obesity.

The impact of psychopathology, social adversity and stress-relevant DNA methylation on prospective risk for post-traumatic stress: A machine learning approach

BACKGROUND

A range of factors have been identified that contribute to greater incidence, severity, and prolonged course of post-traumatic stress disorder (PTSD), including: comorbid and/or prior psychopathology; social adversity such as low socioeconomic position, perceived discrimination, and isolation; and biological factors such as genomic variation at glucocorticoid receptor regulatory network (GRRN) genes. This complex etiology and clinical course make identification of people at higher risk of PTSD challenging. Here we leverage machine learning (ML) approaches to identify a core set of factors that may together predispose persons to PTSD.

METHODS

We used multiple ML approaches to assess the relationship among DNA methylation (DNAm) at GRRN genes, prior psychopathology, social adversity, and prospective risk for PTS severity (PTSS).

RESULTS

ML models predicted prospective risk of PTSS with high accuracy. The Gradient Boost approach was the top-performing model with mean absolute error of 0.135, mean square error of 0.047, root mean square error of 0.217, and R2 of 95.29%. Prior PTSS ranked highest in predicting the prospective risk of PTSS, accounting for >88% of the prediction. The top ranked GRRN CpG site was cg05616442, in AKT1, and the top ranked social adversity feature was loneliness.

CONCLUSION

Multiple factors including prior PTSS, social adversity, and DNAm play a role in predicting prospective risk of PTSS. ML models identified factors accounting for increased PTSS risk with high accuracy, which may help to target risk factors that reduce the likelihood or course of PTSD, potentially pointing to approaches that can lead to early intervention.

LIMITATION

One of the limitations of this study is small sample size.

DNA Methylation in Peripheral Blood and Risk of Gastric Cancer: A Prospective Nested Case-control Study

DNA methylation in peripheral blood is a potential biomarker of gastric cancer risk which could be used for early detection. We conducted a prospective case-control study nested within the Melbourne Collaborative Cohort Study. Genomic DNA was prepared from blood samples collected a median of 12 years before diagnosis for cases (N = 168). Controls (N = 163) were matched to cases on sex, year of birth, country of birth, and blood sample type using incidence density sampling. Genome-wide DNA methylation was measured using the Infinium HumanMethylation450K Beadchip. Global measures of DNA methylation were defined as the median methylation M value, calculated for each of 13 CpG subsets representing genomic function, mean methylation and location, and reliability of measurement. Conditional logistic regression was conducted to assess associations between these global measures of methylation and gastric cancer risk, adjusting for Helicobacter pylori and other potential confounders. We tested nonlinear associations using quintiles of the global measure distribution. A genome-wide association study of DNA methylation and gastric cancer risk was also conducted (N = 484,989 CpGs) using conditional logistic regression, adjusting for potential confounders. Differentially methylated regions (DMR) were investigated using the R package DMRcate We found no evidence of associations with gastric cancer risk for individual CpGs or DMRs (P > 7.6 × 10^-6). No evidence of association was observed with global measures of methylation (OR 1.07 per SD of overall median methylation; 95% confidence interval, 0.80-1.44; P = 0.65). We found no evidence that blood DNA methylation is prospectively associated with gastric cancer risk.Prevention Relevance: We studied DNA methylation in blood to try and predict who was at risk of gastric cancer before symptoms developed, by which stage survival is poor. We did not find any such markers, but the importance of early diagnosis in gastric cancer remains, and the search for markers continues.

Simulating ComBat: how batch correction can lead to the systematic introduction of false positive results in DNA methylation microarray studies

Background

Systematic technical effects—also called batch effects—are a considerable challenge when analyzing DNA methylation (DNAm) microarray data, because they can lead to false results when confounded with the variable of interest. Methods to correct these batch effects are error-prone, as previous findings have shown.

Results

Here, we demonstrate how using the R function ComBat to correct simulated Infinium HumanMethylation450 BeadChip (450 K) and Infinium MethylationEPIC BeadChip Kit (EPIC) DNAm data can lead to a large number of false positive results under certain conditions. We further provide a detailed assessment of the consequences for the highly relevant problem of p-value inflation with subsequent false positive findings after application of the frequently used ComBat method. Using ComBat to correct for batch effects in randomly generated samples produced alarming numbers of false discovery rate (FDR) and Bonferroni-corrected (BF) false positive results in unbalanced as well as in balanced sample distributions in terms of the relation between the outcome of interest variable and the technical position of the sample during the probe measurement. Both sample size and number of batch factors (e.g. number of chips) were systematically simulated to assess the probability of false positive findings. The effect of sample size was simulated using n = 48 up to n = 768 randomly generated samples. Increasing the number of corrected factors led to an exponential increase in the number of false positive signals. Increasing the number of samples reduced, but did not completely prevent, this effect.

Conclusions

Using the approach described, we demonstrate, that using ComBat for batch correction in DNAm data can lead to false positive results under certain conditions and sample distributions. Our results are thus contrary to previous publications, considering a balanced sample distribution as unproblematic when using ComBat. We do not claim completeness in terms of reporting all technical conditions and possible solutions of the occurring problems as we approach the problem from a clinician’s perspective and not from that of a computer scientist. With our approach of simulating data, we provide readers with a simple method to assess the probability of false positive findings in DNAm microarray data analysis pipelines.

EWAS Data Hub: a resource of DNA methylation array data and metadata

Epigenome-Wide Association Study (EWAS) has become an effective strategy to explore epigenetic basis of complex traits. Over the past decade, a large amount of epigenetic data, especially those sourced from DNA methylation array, has been accumulated as the result of numerous EWAS projects. We present EWAS Data Hub (https://bigd.big.ac.cn/ewas/datahub), a resource for collecting and normalizing DNA methylation array data as well as archiving associated metadata. The current release of EWAS Data Hub integrates a comprehensive collection of DNA methylation array data from 75 344 samples and employs an effective normalization method to remove batch effects among different datasets. Accordingly, taking advantages of both massive high-quality DNA methylation data and standardized metadata, EWAS Data Hub provides reference DNA methylation profiles under different contexts, involving 81 tissues/cell types (that contain 25 brain parts and 25 blood cell types), six ancestry categories, and 67 diseases (including 39 cancers). In summary, EWAS Data Hub bears great promise to aid the retrieval and discovery of methylation-based biomarkers for phenotype characterization, clinical treatment and health care.

The tissue-specific aspect of genome-wide DNA methylation in newborn and placental tissues: implications for epigenetic epidemiologic studies

Epigenetic programming is essential for lineage differentiation, embryogenesis and placentation in early pregnancy. In epigenetic association studies, DNA methylation is often examined in DNA derived from white blood cells, although its validity to other tissues of interest remains questionable. Therefore, we investigated the tissue specificity of epigenome-wide DNA methylation in newborn and placental tissues. Umbilical cord white blood cells (UC-WBC, n = 25), umbilical cord blood mononuclear cells (UC-MNC, n = 10), human umbilical vein endothelial cells (HUVEC, n = 25) and placental tissue (n = 25) were obtained from 36 uncomplicated pregnancies. Genome-wide DNA methylation was measured by the Illumina HumanMethylation450K BeadChip. Using UC-WBC as a reference tissue, we identified 3595 HUVEC tissue-specific differentially methylated regions (tDMRs) and 11,938 placental tDMRs. Functional enrichment analysis showed that HUVEC and placental tDMRs were involved in embryogenesis, vascular development and regulation of gene expression. No tDMRs were identified in UC-MNC. In conclusion, the extensive amount of genome-wide HUVEC and placental tDMRs underlines the relevance of tissue-specific approaches in future epigenetic association studies, or the use of validated representative tissues for a certain disease of interest, if available. To this purpose, we herewith provide a relevant dataset of paired, tissue-specific, genome-wide methylation measurements in newborn tissues.

RANDOMIZE: A Web Server for Data Randomization

The microarray-based Illumina Infinium MethylationEpic BeadChip (Epic 850k) has become a useful and standard tool for epigenome wide deoxyribonucleic acid (DNA) methylation profiling. Data from this technology may suffer from batch effects due to improper handling of the samples during the plating process. Batch effects are a significant issue and can give rise to spurious and inaccurate results and reduction in power to detect real biological differences. Careful study design, such as randomizing the samples to uniformly distribute the samples across the factors responsible for batch effects, is crucial to address batch effects and other technical artifacts. Randomization helps to reduce the likelihood of bias and impact of difference among groups. This process of randomizing the samples can be a tedious, error-prone, and time-consuming task without a user-friendly and efficient tool. We present RANDOMIZE, a web-based application designed to perform randomization of relevant metadata to evenly distribute samples across the factors typically responsible for batch effects in DNA methylation microarrays, such as rows, chips and plates. We demonstrate that the tool is efficient, fast and easy to use. The tool is freely available online at https://coph-usf.shinyapps.io/RANDOMIZE/ and can be accessed using any web browser. Sample data and tutorial is also available with the tool.

MBD-seq - realities of a misunderstood method for high-quality methylome-wide association studies

The majority of methylome-wide association studies (MWAS) have been performed using commercially available array-based technologies such as the Infinium Human Methylation 450K and the Infinium MethylationEPIC arrays (Illumina). While these arrays offer a convenient and relatively robust assessment of the probed sites they only allow interrogation of 2-4% of all CpG sites in the human genome. Methyl-binding domain sequencing (MBD-seq) is an alternative approach for MWAS that provides near-complete coverage of the methylome at similar costs as the array-based technologies. However, despite publication of multiple positive evaluations, the use of MBD-seq for MWAS is often fiercely criticized. Here we discuss key features of the method and debunk misconceptions using empirical data. We conclude that MBD-seq represents an excellent approach for large-scale MWAS and that increased utilization is likely to result in more discoveries, advance biological knowledge, and expedite the clinical translation of methylome-wide research findings.

Gene set analysis methods for the functional interpretation of non-mRNA data—Genomic range and ncRNA data

Gene set analysis (GSA) is one of the methods of choice for analyzing the results of current omics studies; however, it has been mainly developed to analyze mRNA (microarray, RNA-Seq) data. The following review includes an update regarding general methods and resources for GSA and then emphasizes GSA methods and tools for non-mRNA omics datasets, specifically genomic range data (ChIP-Seq, SNP and methylation) and ncRNA data (miRNAs, lncRNAs and others). In the end, the state of the GSA field for non-mRNA datasets is discussed, and some current challenges and trends are highlighted, especially the use of network approaches to face complexity issues.

Don't brush off buccal data heterogeneity

Buccal epithelial cells are among the most clinically accessible tissues and are increasingly being used to identify epigenetic disease patterns. However, substantial variation in buccal DNA methylation patterns indicates heterogeneity of cell types within and between samples, raising questions of data quality. We systematically estimated cell-type composition for a large collection of buccal and saliva samples from 11 published studies of DNA methylation. In these we identified numerous cases of buccal samples with questionable purity, which may be affected by sampling from individuals with neurodevelopmental disorders, and by the brushes used for sample collection. Further challenges are involved in comparisons with tissues such as saliva, in which buccal component varies widely. We propose a reference-based method of correcting for buccal purity that reduces unwanted variation while preserving cross-tissue differences. Our work demonstrates the wide variation of buccal quality in epigenetic studies and suggests a possible approach to overcome this issue.

DNA methylation in the human frontal cortex reveals a putative mechanism for age-by-disease interactions

A consistent gene set undergoes age-associated expression changes in the human cerebral cortex, and our Age-by-Disease Model posits that these changes contribute to psychiatric diseases by "pushing" the expression of disease-associated genes in disease-promoting directions. DNA methylation (DNAm) is an attractive candidate mechanism for age-associated gene expression changes. We used the Illumina HumanMethylation450 array to characterize genome-wide DNAm in the postmortem orbital frontal cortex from 20 younger (<42 years) and 19 older (>60 years) subjects. DNAm data were integrated with existing normal brain aging expression data and sets of psychiatric disease risk genes to test the hypothesis that age-associated DNAm changes contribute to age-associated gene expression changes and, by extension, susceptibility to psychiatric diseases. We found that age-associated differentially methylated regions (aDMRs) are common, robust, bidirectional, concentrated in CpG island shelves and sea, depleted in CpG islands, and enriched among genes undergoing age-associated expression changes (OR = 2.30, p = 1.69 × 10^-27). We found the aDMRs are enriched among genetic association-based risk genes for schizophrenia, Alzheimer's disease (AD), and major depressive disorder (MDD) (OR = 2.51, p = 0.00015; OR = 2.38, p = 0.036; and OR = 3.08, p = 0.018, respectively) as well as expression-based MDD-associated genes (OR = 1.48, p = 0.00012). Similar patterns of enrichment were found for aDMRs that correlate with local gene expression. These results were replicated in a large publically-available dataset, and confirmed by meta-analysis of the two datasets. Our findings suggest DNAm is a molecular mechanism for age-associated gene expression changes and support a role for DNAm in age-by-disease interactions through preferential targeting of disease-associated genes.

Methylation Analysis Using Microarrays: Analysis and Interpretation

This chapter discusses analysis and interpretation of large-scale Illumina DNA methylation microarray data, used in the context of cancer studies. We outline commonly used normalization procedures and list issues to consider regarding data preprocessing. Focusing on software packages for R, we describe methods for finding features in the methylation data that are of importance for generating and testing hypotheses in cancer research, like differentially methylated positions or regions and global methylation trends.

Heritable methylation marks associated with breast and prostate cancer risk

BACKGROUND

DNA methylation can mimic the effects of germline mutations in cancer predisposition genes. Recently, we identified twenty-four heritable methylation marks associated with breast cancer risk. As breast and prostate cancer share genetic risk factors, including rare, high-risk mutations (eg, in BRCA2), we hypothesized that some of these heritable methylation marks might also be associated with the risk of prostate cancer.

METHODS

We studied 869 incident prostate cancers (430 aggressive and 439 non-aggressive) and 869 matched controls nested within a prospective cohort study. DNA methylation was measured in pre-diagnostic blood samples using the Illumina Infinium HM450K BeadChip. Conditional logistic regression models, adjusted for prostate cancer risk factors and blood cell composition, were used to estimate odds ratios and 95% confidence intervals for the association between the 24 methylation marks and the risk of prostate cancer.

RESULTS

Five methylation marks within the VTRNA2-1 promoter region (cg06536614, cg00124993, cg26328633, cg25340688, and cg26896946), and one in the body of CLGN (cg22901919) were associated with the risk of prostate cancer. In stratified analyses, the five VTRNA2-1 marks were associated with the risk of aggressive prostate cancer.

CONCLUSIONS

This work highlights a potentially important new area of investigation for prostate cancer susceptibility and adds to our knowledge about shared risk factors for breast and prostate cancer.

An epigenome-wide study of cord blood DNA methylations in relation to prenatal perfluoroalkyl substance exposure: The Hokkaido study

BACKGROUND

Prenatal exposure to perfluoroalkyl substances (PFASs) influences fetal development and later in life.

OBJECTIVE

To investigate cord blood DNA methylation changes associated with prenatal exposure to PFASs.

METHODS

We assessed DNA methylation in cord blood samples from 190 mother-child pairs from the Sapporo cohort of the Hokkaido Study (discovery cohort) and from 37 mother-child pairs from the Taiwan Maternal and Infant Cohort Study (replication cohort) using the Illumina HumanMethylation 450 BeadChip. We examined the associations between methylation and PFAS levels in maternal serum using robust linear regression models and identified differentially methylated positions (DMPs) and regions (DMRs).

RESULTS

We found four DMPs with a false discovery rate below 0.05 in the discovery cohort. Among the top 20 DMPs ranked by the lowest P-values for perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) exposure, four DMPs showed the same direction of effect and P-value < 0.05 in the replication assay: cg16242615 mapped to ZBTB7A, cg21876869 located in the intergenic region (IGR) of USP2-AS1, cg00173435 mapped to TCP11L2, and cg18901140 located in IGR of NTN1. For DMRs, we found a region associated with PFOA exposure with family-wise error rate < 0.1 located in ZFP57, showing the same direction of effect in the replication cohort. Among the top five DMRs ranked by the lowest P-values that were associated with exposure to PFOS and PFOA, in addition to ZFP57, DMRs in the CYP2E1, SMAD3, SLC17A9, GFPT2, DUSP22, and TCERG1L genes showed the same direction of effect in the replication cohort.

CONCLUSION

We suggest that prenatal exposure to PFASs may affect DNA methylation status at birth. Longitudinal studies are needed to examine whether methylation changes observed are associated with differential health outcomes.

A Low Glycaemic Index Diet in Pregnancy Induces DNA Methylation Variation in Blood of Newborns: Results from the ROLO Randomised Controlled Trial

The epigenetic profile of the developing fetus is sensitive to environmental influence. Maternal diet has been shown to influence DNA methylation patterns in offspring, but research in humans is limited. We investigated the impact of a low glycaemic index dietary intervention during pregnancy on offspring DNA methylation patterns using a genome-wide methylation approach. Sixty neonates were selected from the ROLO (Randomised cOntrol trial of LOw glycaemic index diet to prevent macrosomia) study: 30 neonates from the low glycaemic index intervention arm and 30 from the control, whose mothers received no specific dietary advice. DNA methylation was investigated in 771,484 CpG sites in free DNA from cord blood serum. Principal component analysis and linear regression were carried out comparing the intervention and control groups. Gene clustering and pathway analysis were also explored. Widespread variation was identified in the newborns exposed to the dietary intervention, accounting for 11% of the total level of DNA methylation variation within the dataset. No association was found with maternal early-pregnancy body mass index (BMI), infant sex, or birthweight. Pathway analysis identified common influences of the intervention on gene clusters plausibly linked to pathways targeted by the intervention, including cardiac and immune functioning. Analysis in 60 additional samples from the ROLO study failed to replicate the original findings. Using a modest-sized discovery sample, we identified preliminary evidence of differential methylation in progeny of mothers exposed to a dietary intervention during pregnancy.

Adjusting for Batch Effects in DNA Methylation Microarray Data, a Lesson Learned

It is well-known, but frequently overlooked, that low- and high-throughput molecular data may contain batch effects, i.e., systematic technical variation. Confounding of experimental batches with the variable(s) of interest is especially concerning, as a batch effect may then be interpreted as a biologically significant finding. An integral step toward reducing false discovery in molecular data analysis includes inspection for batch effects and accounting for this signal if present. In a 30-sample pilot Illumina Infinium HumanMethylation450 (450k array) experiment, we identified two sources of batch effects: row and chip. Here, we demonstrate two approaches taken to process the 450k data in which an R function, ComBat, was applied to adjust for the non-biological signal. In the "initial analysis," the application of ComBat to an unbalanced study design resulted in 9,612 and 19,214 significant (FDR < 0.05) DNA methylation differences, despite none present prior to correction. Suspicious of this dramatic change, a "revised processing" included changes to our analysis as well as a greater number of samples, and successfully reduced batch effects without introducing false signal. Our work supports conclusions made by an article previously published in this journal: though the ultimate antidote to batch effects is thoughtful study design, every DNA methylation microarray analysis should inspect, assess and, if necessary, account for batch effects. The analysis experience presented here can serve as a reminder to the broader community to establish research questions a priori, ensure that they match with study design and encourage communication between technicians and analysts.

Using the Connectivity Map to discover compounds influencing human osteoblast differentiation

Osteoporosis is a common skeletal disorder characterized by low bone mass leading to increased bone fragility and fracture susceptibility. Identification of factors influencing osteoblast differentiation and bone formation is very important. Previously, we identified parbendazole to be a novel compound that stimulates osteogenic differentiation of human mesenchymal stromal cells (hMSCs), using gene expression profiling and bioinformatic analyzes, including the Connectivity Map (CMap), as an in-silico approach. The aim for this paper is to identify additional compounds affecting osteoblast differentiation using the CMap. Gene expression profiling was performed on hMSCs differentiated to osteoblasts using Illumina microarrays. Our osteoblast gene signature, the top regulated genes 6 hr after induction by dexamethasone, was uploaded into CMap (www.broadinstitute.org/cmap/). Through this approach we identified compounds with gene signatures positively correlating (withaferin-A, calcium folinate, amylocaine) or negatively correlating (salbutamol, metaraminol, diprophylline) to our osteoblast gene signature. All positively correlating compounds stimulated osteogenic differentiation, as indicated by increased mineralization compared to control treated cells. One of three negatively correlating compounds, salbutamol, inhibited dexamethasone-induced osteoblastic differentiation, while the other two had no effect. Based on gene expression data of withaferin-A and salbutamol, we identified HMOX1 and STC1 as being strongly differentially expressed . shRNA knockdown of HMOX1 or STC1 in hMSCs inhibited osteoblast differentiation. These results confirm that the CMap is a powerful approach to identify positively compounds that stimulate osteogenesis of hMSCs, and through this approach we can identify genes that play an important role in osteoblast differentiation and could be targets for novel bone anabolic therapies.

DNA methylation-based biological aging and cancer risk and survival: Pooled analysis of seven prospective studies

The association between aging and cancer is complex. Recent studies have developed measures of biological aging based on DNA methylation and called them "age acceleration." We aimed to assess the associations of age acceleration with risk of and survival from seven common cancers. Seven case-control studies of DNA methylation and colorectal, gastric, kidney, lung, prostate and urothelial cancer and B-cell lymphoma nested in the Melbourne Collaborative Cohort Study were conducted. Cancer cases, vital status and cause of death were ascertained through linkage with cancer and death registries. Conditional logistic regression and Cox models were used to estimate odds ratios (OR) and hazard ratios (HR) and 95% confidence intervals (CI) for associations of five age acceleration measures derived from the Human Methylation 450 K Beadchip assay with cancer risk (N = 3,216 cases) and survival (N = 1,726 deaths), respectively. Epigenetic aging was associated with increased cancer risk, ranging from 4% to 9% per five-year age acceleration for the 5 measures considered. Heterogeneity by study was observed, with stronger associations for risk of kidney cancer and B-cell lymphoma. An associated increased risk of death following cancer diagnosis ranged from 2% to 6% per five-year age acceleration, with no evidence of heterogeneity by cancer site. Cancer risk and mortality were increased by 15-30% for the fourth versus first quartile of age acceleration. DNA methylation-based measures of biological aging are associated with increased cancer risk and shorter cancer survival, independently of major health risk factors.

Orienting the causal relationship between imprecisely measured traits using GWAS summary data

Inference about the causal structure that induces correlations between two traits can be achieved by combining genetic associations with a mediation-based approach, as is done in the causal inference test (CIT). However, we show that measurement error in the phenotypes can lead to the CIT inferring the wrong causal direction, and that increasing sample sizes has the adverse effect of increasing confidence in the wrong answer. This problem is likely to be general to other mediation-based approaches. Here we introduce an extension to Mendelian randomisation, a method that uses genetic associations in an instrumentation framework, that enables inference of the causal direction between traits, with some advantages. First, it can be performed using only summary level data from genome-wide association studies; second, it is less susceptible to bias in the presence of measurement error or unmeasured confounding. We apply the method to infer the causal direction between DNA methylation and gene expression levels. Our results demonstrate that, in general, DNA methylation is more likely to be the causal factor, but this result is highly susceptible to bias induced by systematic differences in measurement error between the platforms, and by horizontal pleiotropy. We emphasise that, where possible, implementing MR and appropriate sensitivity analyses alongside other approaches such as CIT is important to triangulate reliable conclusions about causality.

Understanding the causal relationships between pairs of traits is crucial for unravelling the causes of disease. To this end, results from genome-wide association studies are valuable because if a trait is known to be influenced by a genetic variant then this knowledge can be used to test the trait’s causal influences on other traits and diseases. Here we discuss scenarios where the nature of the genetic association with the causal trait can lead existing causal inference methods to give the wrong direction of causality. We introduce a new method that can be applied to summary level data and is potentially less susceptible to problems such as measurement error, and apply it to evaluate the causal relationships between DNA methylation levels and gene expression. While our results show that DNA methylation is more likely to be the causal factor, we point out that is it crucial to acknowledge that systematic differences in measurement error between the platforms could influence such conclusions.

Epigenome-wide association of PTSD from heterogeneous cohorts with a common multi-site analysis pipeline

Compelling evidence suggests that epigenetic mechanisms such as DNA methylation play a role in stress regulation and in the etiologic basis of stress related disorders such as Post traumatic Stress Disorder (PTSD). Here we describe the purpose and methods of an international consortium that was developed to study the role of epigenetics in PTSD. Inspired by the approach used in the Psychiatric Genomics Consortium, we brought together investigators representing seven cohorts with a collective sample size of N = 1147 that included detailed information on trauma exposure, PTSD symptoms, and genome-wide DNA methylation data. The objective of this consortium is to increase the analytical sample size by pooling data and combining expertise so that DNA methylation patterns associated with PTSD can be identified. Several quality control and analytical pipelines were evaluated for their control of genomic inflation and technical artifacts with a joint analysis procedure established to derive comparable data over the cohorts for meta-analysis. We propose methods to deal with ancestry population stratification and type I error inflation and discuss the advantages and disadvantages of applying robust error estimates. To evaluate our pipeline, we report results from an epigenome-wide association study (EWAS) of age, which is a well-characterized phenotype with known epigenetic associations. Overall, while EWAS are highly complex and subject to similar challenges as genome-wide association studies (GWAS), we demonstrate that an epigenetic meta-analysis with a relatively modest sample size can be well-powered to identify epigenetic associations. Our pipeline can be used as a framework for consortium efforts for EWAS.

Epigenetic Signatures of Salivary Gland Inflammation in Sjögren's Syndrome

Objective

Sjögren's syndrome (SS) is a complex multisystem autoimmune disease that results in progressive destruction of the exocrine glands. The purpose of this study was to characterize epigenetic changes in affected gland tissue and describe the relationship of these changes to known inflammatory processes.

Methods

A genome‐wide DNA methylation study was performed on human labial salivary gland (LSG) biopsy samples obtained from 28 female members of the Sjögren's International Collaborative Clinical Alliance (SICCA) Registry. Gland tissue was methylotyped using the Illumina HumanMethylation450 BeadChip platform, followed by rigorous probe‐filtering and data‐normalization procedures.

Results

A genome‐wide case–control study of 26 of the 28 subjects revealed 7,820 differentially methylated positions (DMPs) associated with disease status, including 5,699 hypomethylated and 2,121 hypermethylated DMPs. Further analysis identified 57 genes that were enriched for DMPs in their respective promoters; many are involved in immune response, including 2 previously established SS genetic risk loci. Bioinformatics analysis highlighted an extended region of hypomethylation surrounding PSMB8 and TAP1, consistent with an increased frequency of antigen‐presenting cells in LSG tissue from the SS cases. Transcription factor motif enrichment analysis revealed the specific nature of the genome‐wide methylation differences, demonstrating colocalization of SS‐associated DMPs with stress‐ and immune response–related motifs.

Conclusion

Our findings underscore the utility of CpG methylotyping as an independent probe of active disease processes in SS, offering unique insights into the composition of disease‐relevant tissue. Methylation profiling implicated several genes and pathways previously thought to be involved in disease‐related processes, as well as a number of new candidates.

Rapid and affordable genome-wide bisulfite DNA sequencing by XmaI-reduced representation bisulfite sequencing

Aim: To develop a reduced representation bisulfite sequencing (RRBS) approach for rapid and affordable genome-wide DNA methylation analysis. Methods: We have selected restriction endonuclease XmaI to produce RRBS library fragments. After digestion and partial fill-in DNA fragments were ligated to barcoded adapters, bisulfite converted, size-selected, and sequenced on the Ion Torrent Personal Genome Machine. XmaI-RRBS results were compared with the previously published RRBS data. Results: We have developed an XmaI-RRBS method for rapid and affordable genome-wide DNA methylation analysis, with library preparation taking only 4 days and sequencing possible within 4 h. We have also addressed several challenges in order to further improve the RRBS technology. XmaI-RRBS may be performed on degraded DNA samples and is compatible with the bench-top next-generation sequencing machines.

Genome-Wide Measures of Peripheral Blood Dna Methylation and Prostate Cancer Risk in a Prospective Nested Case-Control Study

BACKGROUND

Global measures of peripheral blood DNA methylation have been associated with risk of some malignancies, including breast, bladder, and gastric cancer. Here, we examined genome-wide measures of peripheral blood DNA methylation in prostate cancer and its non-aggressive and aggressive disease forms.

METHODS

We used a matched, case-control study of 687 incident prostate cancer samples, nested within a larger prospective cohort study. DNA methylation was measured in pre-diagnostic, peripheral blood samples using the Illumina Infinium HM450K BeadChip. Genome-wide measures of DNA methylation were computed as the median M-value of all CpG sites and according to CpG site location and regulatory function. We used conditional logistic regression to test for associations between genome-wide measures of DNA methylation and risk of prostate cancer and its subtypes, and by time between blood draw and diagnosis.

RESULTS

We observed no associations between the genome-wide measure of DNA methylation based on all CpG sites and risk of prostate cancer or aggressive disease. Risk of non-aggressive disease was associated with higher methylation of CpG islands (OR = 0.80; 95%CI = 0.68-0.94), promoter regions (OR = 0.79; 95%CI = 0.66-0.93), and high density CpG regions (OR = 0.80; 95%CI = 0.68-0.94). Additionally, higher methylation of all CpGs (OR = 0.66; 95%CI = 0.48-0.89), CpG shores (OR = 0.62; 95%CI = 0.45-0.84), and regulatory regions (OR = 0.68; 95% CI = 0.51-0.91) was associated with a reduced risk of overall prostate cancer within 5 years of blood draw but not thereafter.

CONCLUSIONS

A reduced risk of overall prostate cancer within 5 years of blood draw and non-aggressive prostate cancer was associated with higher genome-wide methylation of peripheral blood DNA. While these data have no immediate clinical utility, with further work they may provide insight into the early events of prostate carcinogenesis. Prostate 77:471-478, 2017. © 2017 Wiley Periodicals, Inc.

Arsenic exposure from drinking water is associated with decreased gene expression and increased DNA methylation in peripheral blood

BACKGROUND

Exposure to inorganic arsenic increases the risk of cancer and non-malignant diseases. Inefficient arsenic metabolism is a marker for susceptibility to arsenic toxicity. Arsenic may alter gene expression, possibly by altering DNA methylation.

OBJECTIVES

To elucidate the associations between arsenic exposure, gene expression, and DNA methylation in peripheral blood, and the modifying effects of arsenic metabolism.

METHODS

The study participants, women from the Andes, Argentina, were exposed to arsenic via drinking water. Arsenic exposure was assessed as the sum of arsenic metabolites in urine (U-As), using high performance liquid-chromatography hydride-generation inductively-coupled-plasma-mass-spectrometry, and arsenic metabolism efficiency was assessed by the urinary fractions (%) of the individual metabolites. Genome-wide gene expression (N=80 women) and DNA methylation (N=93; 80 overlapping with gene expression) in peripheral blood were measured using Illumina DirectHyb HumanHT-12 v4.0 and Infinium Human-Methylation 450K BeadChip, respectively.

RESULTS

U-As concentrations, ranging 10-1251μg/L, was associated with decreased gene expression: 64% of the top 1000 differentially expressed genes were down-regulated with increasing U-As. U-As was also associated with hypermethylation: 87% of the top 1000CpGs were hypermethylated with increasing U-As. The expression of six genes and six individual CpG sites were significantly associated with increased U-As concentration. Pathway analyses revealed enrichment of genes related to cell death and cancer. The pathways differed somewhat depending on arsenic metabolism efficiency. We found no overlap between arsenic-related gene expression and DNA methylation for individual genes.

CONCLUSIONS

Increased arsenic exposure was associated with lower gene expression and hypermethylation in peripheral blood, but with no evident overlap.

Differential DNA methylation and PM2.5 species in a 450K epigenome-wide association study

Although there is growing evidence that exposure to ambient particulate matter is associated with global DNA methylation and gene-specific methylation, little is known regarding epigenome-wide changes in DNA methylation in relation to particles and, especially, particle components. Using the Illumina Infinium HumanMethylation450 BeadChip, we examined the relationship between one-year moving averages of PM_2.5 species (Al, Ca, Cu, Fe, K, Na, Ni, S, Si, V, and Zn) and DNA methylation at 484,613 CpG probes in a longitudinal cohort that included 646 subjects. Bonferroni correction was applied to adjust for multiple comparisons. Bioinformatics analysis of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment was also performed. We observed 20 Bonferroni significant (P-value < 9.4× 10^-9) CpGs for Fe, 8 for Ni, and 1 for V. Particularly, methylation at Schlafen Family Member 11 (SLFN11) cg10911913 was positively associated with measured levels of all 3 species. The SLFN11 gene codes for an interferon-induced protein that inhibits retroviruses and sensitizes cancer cells to DNA-damaging agents. Bioinformatics analysis suggests that gene targets may be relevant to pathways including cancers, signal transduction, and cell growth and death. Ours is the first study to examine the epigenome-wide association between ambient particles species and DNA methylation. We found that long-term exposures to specific components of ambient particle pollution, especially particles emitted during oil combustion, were associated with methylation changes in genes relevant to immune responses. Our findings provide insight into potential biologic mechanisms on an epigenetic level.

Risk-conscious correction of batch effects: maximising information extraction from high-throughput genomic datasets

BACKGROUND

Batch effects are a persistent and pervasive form of measurement noise which undermine the scientific utility of high-throughput genomic datasets. At their most benign, they reduce the power of statistical tests resulting in actual effects going unidentified. At their worst, they constitute confounds and render datasets useless. Attempting to remove batch effects will result in some of the biologically meaningful component of the measurement (i.e. signal) being lost. We present and benchmark a novel technique, called Harman. Harman maximises the removal of batch noise with the constraint that the risk of also losing biologically meaningful component of the measurement is kept to a fraction which is set by the user.

RESULTS

Analyses of three independent publically available datasets reveal that Harman removes more batch noise and preserves more signal at the same time, than the current leading technique. Results also show that Harman is able to identify and remove batch effects no matter what their relative size compared to other sources of variation in the dataset. Of particular advantage for meta-analyses and data integration is Harman's superior consistency in achieving comparable noise suppression - signal preservation trade-offs across multiple datasets, with differing number of treatments, replicates and processing batches.

CONCLUSION

Harman's ability to better remove batch noise, and better preserve biologically meaningful signal simultaneously within a single study, and maintain the user-set trade-off between batch noise rejection and signal preservation across different studies makes it an effective alternative method to deal with batch effects in high-throughput genomic datasets. Harman is flexible in terms of the data types it can process. It is available publically as an R package ( https://bioconductor.org/packages/release/bioc/html/Harman.html ), as well as a compiled Matlab package ( http://www.bioinformatics.csiro.au/harman/ ) which does not require a Matlab license to run.

Genome-wide measures of DNA methylation in peripheral blood and the risk of urothelial cell carcinoma: a prospective nested case-control study

Background:

Global DNA methylation has been reported to be associated with urothelial cell carcinoma (UCC) by studies using blood samples collected at diagnosis. Using the Illumina HumanMethylation450 assay, we derived genome-wide measures of blood DNA methylation and assessed them for their prospective association with UCC risk.

Methods:

We used 439 case–control pairs from the Melbourne Collaborative Cohort Study matched on age, sex, country of birth, DNA sample type, and collection period. Conditional logistic regression was used to compute odds ratios (OR) of UCC risk per s.d. of each genome-wide measure of DNA methylation and 95% confidence intervals (CIs), adjusted for potential confounders. We also investigated associations by disease subtype, sex, smoking, and time since blood collection.

Results:

The risk of superficial UCC was decreased for individuals with higher levels of our genome-wide DNA methylation measure (OR=0.71, 95% CI: 0.54–0.94; P=0.02). This association was particularly strong for current smokers at sample collection (OR=0.47, 95% CI: 0.27–0.83). Intermediate levels of our genome-wide measure were associated with decreased risk of invasive UCC. Some variation was observed between UCC subtypes and the location and regulatory function of the CpGs included in the genome-wide measures of methylation.

Conclusions:

Higher levels of our genome-wide DNA methylation measure were associated with decreased risk of superficial UCC and intermediate levels were associated with reduced risk of invasive disease. These findings require replication by other prospective studies.

Genome-wide characteristics of de novo mutations in autism

De novo mutations (DNMs) are important in Autism Spectrum Disorder (ASD), but so far analyses have mainly been on the ~1.5% of the genome encoding genes. Here, we performed whole genome sequencing (WGS) of 200 ASD parent-child trios and characterized germline and somatic DNMs. We confirmed that the majority of germline DNMs (75.6%) originated from the father, and these increased significantly with paternal age only (p=4.2×10^-10). However, when clustered DNMs (those within 20kb) were found in ASD, not only did they mostly originate from the mother (p=7.7×10^-13), but they could also be found adjacent to de novo copy number variations (CNVs) where the mutation rate was significantly elevated (p=2.4×10^-24). By comparing DNMs detected in controls, we found a significant enrichment of predicted damaging DNMs in ASD cases (p=8.0×10^-9; OR=1.84), of which 15.6% (p=4.3×10^-3) and 22.5% (p=7.0×10^-5) were in the non-coding or genic non-coding, respectively. The non-coding elements most enriched for DNM were untranslated regions of genes, boundaries involved in exon-skipping and DNase I hypersensitive regions. Using microarrays and a novel outlier detection test, we also found aberrant methylation profiles in 2/185 (1.1%) of ASD cases. These same individuals carried independently identified DNMs in the ASD risk- and epigenetic- genes DNMT3A and ADNP. Our data begins to characterize different genome-wide DNMs, and highlight the contribution of non-coding variants, to the etiology of ASD.

Tobacco smoking-associated genome-wide DNA methylation changes in the EPIC study

AIM

Epigenetic changes may occur in response to environmental stressors, and an altered epigenome pattern may represent a stable signature of environmental exposure.

MATERIALS & METHODS

Here, we examined the potential of DNA methylation changes in 910 prediagnostic peripheral blood samples as a marker of exposure to tobacco smoke in a large multinational cohort.

RESULTS

We identified 748 CpG sites that were differentially methylated between smokers and nonsmokers, among which we identified novel regionally clustered CpGs associated with active smoking. Importantly, we found a marked reversibility of methylation changes after smoking cessation, although specific genes remained differentially methylated up to 22 years after cessation.

CONCLUSION

Our study has comprehensively cataloged the smoking-associated DNA methylation alterations and showed that these alterations are reversible after smoking cessation.

Establishing an analytic pipeline for genome-wide DNA methylation

The need for research investigating DNA methylation (DNAm) in clinical studies has increased, leading to the evolution of new analytic methods to improve accuracy and reproducibility of the interpretation of results from these studies. The purpose of this article is to provide clinical researchers with a summary of the major data processing steps routinely applied in clinical studies investigating genome-wide DNAm using the Illumina HumanMethylation 450K BeadChip. In most studies, the primary goal of employing DNAm analysis is to identify differential methylation at CpG sites among phenotypic groups. Experimental design considerations are crucial at the onset to minimize bias from factors related to sample processing and avoid confounding experimental variables with non-biological batch effects. Although there are currently no de facto standard methods for analyzing these data, we review the major steps in processing DNAm data recommended by several research studies. We describe several variations available for clinical researchers to process, analyze, and interpret DNAm data. These insights are applicable to most types of genome-wide DNAm array platforms and will be applicable for the next generation of DNAm array technologies (e.g., the 850K array). Selection of the DNAm analytic pipeline followed by investigators should be guided by the research question and supported by recently published methods.

Improving cell mixture deconvolution by identifying optimal DNA methylation libraries (IDOL)

Background

Confounding due to cellular heterogeneity represents one of the foremost challenges currently facing Epigenome-Wide Association Studies (EWAS). Statistical methods leveraging the tissue-specificity of DNA methylation for deconvoluting the cellular mixture of heterogenous biospecimens offer a promising solution, however the performance of such methods depends entirely on the library of methylation markers being used for deconvolution. Here, we introduce a novel algorithm for Identifying Optimal Libraries (IDOL) that dynamically scans a candidate set of cell-specific methylation markers to find libraries that optimize the accuracy of cell fraction estimates obtained from cell mixture deconvolution.

Results

Application of IDOL to training set consisting of samples with both whole-blood DNA methylation data (Illumina HumanMethylation450 BeadArray (HM450)) and flow cytometry measurements of cell composition revealed an optimized library comprised of 300 CpG sites. When compared existing libraries, the library identified by IDOL demonstrated significantly better overall discrimination of the entire immune cell landscape (p = 0.038), and resulted in improved discrimination of 14 out of the 15 pairs of leukocyte subtypes. Estimates of cell composition across the samples in the training set using the IDOL library were highly correlated with their respective flow cytometry measurements, with all cell-specific R²>0.99 and root mean square errors (RMSEs) ranging from [0.97 % to 1.33 %] across leukocyte subtypes. Independent validation of the optimized IDOL library using two additional HM450 data sets showed similarly strong prediction performance, with all cell-specific R²>0.90 and RMSE<4.00 %. In simulation studies, adjustments for cell composition using the IDOL library resulted in uniformly lower false positive rates compared to competing libraries, while also demonstrating an improved capacity to explain epigenome-wide variation in DNA methylation within two large publicly available HM450 data sets.

Conclusions

Despite consisting of half as many CpGs compared to existing libraries for whole blood mixture deconvolution, the optimized IDOL library identified herein resulted in outstanding prediction performance across all considered data sets and demonstrated potential to improve the operating characteristics of EWAS involving adjustments for cell distribution. In addition to providing the EWAS community with an optimized library for whole blood mixture deconvolution, our work establishes a systematic and generalizable framework for the assembly of libraries that improve the accuracy of cell mixture deconvolution.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-0943-7) contains supplementary material, which is available to authorized users.

An epigenome-wide association meta-analysis of prenatal maternal stress in neonates: A model approach for replication

Prenatal maternal stress exposure has been associated with neonatal differential DNA methylation. However, the available evidence in humans is largely based on candidate gene methylation studies, where only a few CpG sites were evaluated. The aim of this study was to examine the association between prenatal exposure to maternal stress and offspring genome-wide cord blood methylation using different methods. First, we conducted a meta-analysis and follow-up pathway analyses. Second, we used novel region discovery methods [i.e., differentially methylated regions (DMRs) analyses]. To this end, we used data from two independent population-based studies, the Generation R Study (n = 912) and the Avon Longitudinal Study of Parents and Children (ALSPAC, n = 828), to (i) measure genome-wide DNA methylation in cord blood and (ii) extract a prenatal maternal stress composite. The meta-analysis (ntotal = 1,740) revealed no epigenome-wide (meta P <1.00e-07) associations of prenatal maternal stress exposure with neonatal differential DNA methylation. Follow-up analyses of the top hits derived from our epigenome-wide meta-analysis (meta P <1.00e-04) indicated an over-representation of the methyltransferase activity pathway. We identified no Bonferroni-corrected (P <1.00e-06) DMRs associated with prenatal maternal stress exposure. Combining data from two independent population-based samples in an epigenome-wide meta-analysis, the current study indicates that there are no large effects of prenatal maternal stress exposure on neonatal DNA methylation. Such replication efforts are essential in the search for robust associations, whether derived from candidate gene methylation or epigenome-wide studies.

Adjustment of Cell-Type Composition Minimizes Systematic Bias in Blood DNA Methylation Profiles Derived by DNA Collection Protocols

Differences in DNA collection protocols may be a potential confounder in epigenome-wide association studies (EWAS) using a large number of blood specimens from multiple biobanks and/or cohorts. Here we show that pre-analytical procedures involved in DNA collection can induce systematic bias in the DNA methylation profiles of blood cells that can be adjusted by cell-type composition variables. In Experiment 1, whole blood from 16 volunteers was collected to examine the effect of a 24 h storage period at 4°C on DNA methylation profiles as measured using the Infinium HumanMethylation450 BeadChip array. Our statistical analysis showed that the P-value distribution of more than 450,000 CpG sites was similar to the theoretical distribution (in quantile-quantile plot, λ = 1.03) when comparing two control replicates, which was remarkably deviated from the theoretical distribution (λ = 1.50) when comparing control and storage conditions. We then considered cell-type composition as a possible cause of the observed bias in DNA methylation profiles and found that the bias associated with the cold storage condition was largely decreased (λ_adjusted = 1.14) by taking into account a cell-type composition variable. As such, we compared four respective sample collection protocols used in large-scale Japanese biobanks or cohorts as well as two control replicates. Systematic biases in DNA methylation profiles were observed between control and three of four protocols without adjustment of cell-type composition (λ = 1.12–1.45) and no remarkable biases were seen after adjusting for cell-type composition in all four protocols (λ_adjusted = 1.00–1.17). These results revealed important implications for comparing DNA methylation profiles between blood specimens from different sources and may lead to discovery of disease-associated DNA methylation markers and the development of DNA methylation profile-based predictive risk models.

Genomic responses to hepatitis B virus (HBV) infection in primary human hepatocytes

Viral infections are able to modify the host's cellular programs, with DNA methylation being a biological intermediate in this process. The extent to which viral infections deregulate gene expression and DNA methylation is not fully understood. In the case of Hepatitis B virus (HBV), there is evidence for an interaction between viral proteins and the host DNA methylation machinery. We studied the ability of HBV to modify the host transcriptome and methylome, using naturally infected primary human hepatocytes to better mimic the clinical setting.Gene expression was especially sensitive to culture conditions, independently of HBV infection. However, we identified non-random changes in gene expression and DNA methylation occurring specifically upon HBV infection. There was little correlation between expression and methylation changes, with transcriptome being a more sensitive marker of time-dependent changes induced by HBV. In contrast, a set of differentially methylated sites appeared early and were stable across the time course experiment. Finally, HBV-induced DNA methylation changes were defined by a specific chromatin context characterized by CpG-poor regions outside of gene promoters.These data support the ability of HBV to modulate host cell expression and methylation programs. In addition, it may serve as a reference for studies addressing the genome-wide consequences of HBV infection in human hepatocytes.

Base resolution methylome profiling: considerations in platform selection, data preprocessing and analysis

Bisulfite treatment-based methylation microarray (mainly Illumina 450K Infinium array) and next-generation sequencing (reduced representation bisulfite sequencing, Agilent SureSelect Human Methyl-Seq, NimbleGen SeqCap Epi CpGiant or whole-genome bisulfite sequencing) are commonly used for base resolution DNA methylome research. Although multiple tools and methods have been developed and used for the data preprocessing and analysis, confusions remains for these platforms including how and whether the 450k array should be normalized; which platform should be used to better fit researchers' needs; and which statistical models would be more appropriate for differential methylation analysis. This review presents the commonly used platforms and compares the pros and cons of each in methylome profiling. We then discuss approaches to study design, data normalization, bias correction and model selection for differentially methylated individual CpGs and regions.

Influence of environmental exposure on human epigenetic regulation

Environmental toxicants can alter epigenetic regulatory features such as DNA methylation and microRNA expression. As the sensitivity of epigenomic regulatory features may be greatest during the in utero period, when critical windows are narrow, and when epigenomic profiles are being set, this review will highlight research focused on that period. I will focus on work in human populations, where the impact of environmental toxicants in utero, including cigarette smoke and toxic trace metals such as arsenic, mercury and manganese, on genome-wide, gene-specific DNA methylation has been assessed. In particular, arsenic is highlighted, as this metalloid has been the focus of a number of studies and its detoxification mechanisms are well understood. Importantly, the tissues and cells being examined must be considered in context in order to interpret the findings of these studies. For example, by studying the placenta, it is possible to identify potential epigenetic adaptations of key genes and pathways that may alter the developmental course in line with the developmental origins of health and disease paradigm. Alternatively, studies of newborn cord blood can be used to examine how environmental exposure in utero can impact the composition of cells within the peripheral blood, leading to immunological effects of exposure. The results suggest that in humans, like other vertebrates, there is a susceptibility for epigenomic alteration by the environment during intrauterine development, and this may represent a mechanism of plasticity of the organism in response to its environment as well as a mechanism through which long-term health consequences can be shaped.

Exposure to aflatoxin B1 in utero is associated with DNA methylation in white blood cells of infants in The Gambia

BACKGROUND

Exposure to environmental toxins during embryonic development may lead to epigenetic changes that influence disease risk in later life. Aflatoxin is a contaminant of staple foods in sub-Saharan Africa, is a known human liver carcinogen and has been associated with stunting in infants.

METHODS

We have measured aflatoxin exposure in 115 pregnant women in The Gambia and examined the DNA methylation status of white blood cells from their infants at 2-8 months old (mean 3.6 ± 0.9). Aflatoxin exposure in women was assessed using an ELISA method to measure aflatoxin albumin (AF-alb) adducts in plasma taken at 1-16 weeks of pregnancy. Genome-wide DNA methylation of infant white blood cells was measured using the Illumina Infinium HumanMethylation450beadchip.

RESULTS

AF-alb levels ranged from 3.9 to 458.4 pg/mg albumin. We found that aflatoxin exposure in the mothers was associated to DNA methylation in their infants for 71 CpG sites (false discovery rate < 0.05), with an average effect size of 1.7% change in methylation. Aflatoxin-associated differential methylation was observed in growth factor genes such as FGF12 and IGF1, and immune-related genes such as CCL28, TLR2 and TGFBI. Moreover, one aflatoxin-associated methylation region (corresponding to the miR-4520b locus) was identified.

CONCLUSIONS

This study shows that maternal exposure to aflatoxin during the early stages of pregnancy is associated with differential DNA methylation patterns of infants, including in genes related to growth and immune function. This reinforces the need for interventions to reduce aflatoxin exposure, especially during critical periods of fetal and infant development.

Removing unwanted variation in a differential methylation analysis of Illumina HumanMethylation450 array data

Due to their relatively low-cost per sample and broad, gene-centric coverage of CpGs across the human genome, Illumina's 450k arrays are widely used in large scale differential methylation studies. However, by their very nature, large studies are particularly susceptible to the effects of unwanted variation. The effects of unwanted variation have been extensively documented in gene expression array studies and numerous methods have been developed to mitigate these effects. However, there has been much less research focused on the appropriate methodology to use for accounting for unwanted variation in methylation array studies. Here we present a novel 2-stage approach using RUV-inverse in a differential methylation analysis of 450k data and show that it outperforms existing methods.

Enzymatic cleavage of histone H3: a new consideration when measuring histone modifications in human samples

Histone modifications are increasingly being used as biomarkers of cancer prognosis and survival. However, we identified a cleavage product of histone H3 in human peripheral blood mononuclear cells, which interferes with measures of certain H3 modifications. Therefore, the potential for enzymatic cleavage of histones should be considered when measuring histone modifications in human samples. Furthermore, the enzymatic cleavage of human H3 is itself a fascinating area of research and two important questions remain to be answered: 1) Does cleavage of human H3 occur in vivo, as it does in other organisms? and 2) Does it serve a biologically important function?