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

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.

Considerations when processing and interpreting genomics data of the placenta

The application of genomic approaches to placental research has opened exciting new avenues to help us understand basic biological properties of the placenta, improve prenatal screening/diagnosis, and measure effects of in utero exposures on child health outcomes. In the last decade, such large-scale genomic data (including epigenomics and transcriptomics) have become more easily accessible to researchers from many disciplines due to the increasing ease of obtaining such data and the rapidly evolving computational tools available for analysis. While the potential of large-scale studies has been widely promoted, less attention has been given to some of the challenges associated with processing and interpreting such data. We hereby share some of our experiences in assessing data quality, reproducibility, and interpretation in the context of genome-wide studies of the placenta, with the aim to improve future studies. There is rarely a single "best" approach, as that can depend on the study question and sample cohort. However, being consistent, thoroughly assessing potential confounders in the data, and communicating key variables in the methods section of the manuscript are critically important to help researchers to collaborate and build on each other's work.

Accurate ethnicity prediction from placental DNA methylation data

Background

The influence of genetics on variation in DNA methylation (DNAme) is well documented. Yet confounding from population stratification is often unaccounted for in DNAme association studies. Existing approaches to address confounding by population stratification using DNAme data may not generalize to populations or tissues outside those in which they were developed. To aid future placental DNAme studies in assessing population stratification, we developed an ethnicity classifier, PlaNET (Placental DNAme Elastic Net Ethnicity Tool), using five cohorts with Infinium Human Methylation 450k BeadChip array (HM450k) data from placental samples that is also compatible with the newer EPIC platform.

Results

Data from 509 placental samples were used to develop PlaNET and show that it accurately predicts (accuracy = 0.938, kappa = 0.823) major classes of self-reported ethnicity/race (African: n = 58, Asian: n = 53, Caucasian: n = 389), and produces ethnicity probabilities that are highly correlated with genetic ancestry inferred from genome-wide SNP arrays (> 2.5 million SNP) and ancestry informative markers (n = 50 SNPs). PlaNET’s ethnicity classification relies on 1860 HM450K microarray sites, and over half of these were linked to nearby genetic polymorphisms (n = 955). Our placental-optimized method outperforms existing approaches in assessing population stratification in placental samples from individuals of Asian, African, and Caucasian ethnicities.

Conclusion

PlaNET provides an improved approach to address population stratification in placental DNAme association studies. The method can be applied to predict ethnicity as a discrete or continuous variable and will be especially useful when self-reported ethnicity information is missing and genotyping markers are unavailable.

Electronic supplementary material

The online version of this article (10.1186/s13072-019-0296-3) contains supplementary material, which is available to authorized users.

Accelerating research on biological aging and mental health: Current challenges and future directions

Aging is associated with complex biological changes that can be accelerated, slowed, or even temporarily reversed by biological and non-biological factors. This article focuses on the link between biological aging, psychological stressors, and mental illness. Rather than comprehensively reviewing this rapidly expanding field, we highlight challenges in this area of research and propose potential strategies to accelerate progress in this field. This effort requires the interaction of scientists across disciplines - including biology, psychiatry, psychology, and epidemiology; and across levels of analysis that emphasize different outcome measures - functional capacity, physiological, cellular, and molecular. Dialogues across disciplines and levels of analysis naturally lead to new opportunities for discovery but also to stimulating challenges. Some important challenges consist of 1) establishing the best objective and predictive biological age indicators or combinations of indicators, 2) identifying the basis for inter-individual differences in the rate of biological aging, and 3) examining to what extent interventions can delay, halt or temporarily reverse aging trajectories. Discovering how psychological states influence biological aging, and vice versa, has the potential to create novel and exciting opportunities for healthcare and possibly yield insights into the fundamental mechanisms that drive human aging.

Placental epigenetic clocks: estimating gestational age using placental DNA methylation levels

The human pan-tissue epigenetic clock is widely used for estimating age across the entire lifespan, but it does not lend itself well to estimating gestational age (GA) based on placental DNAm methylation (DNAm) data. We replicate previous findings demonstrating a strong correlation between GA and genome-wide DNAm changes. Using substantially more DNAm arrays (n=1,102 in the training set) than a previous study, we present three new placental epigenetic clocks: 1) a robust placental clock (RPC) which is unaffected by common pregnancy complications (e.g., gestational diabetes, preeclampsia), and 2) a control placental clock (CPC) constructed using placental samples from pregnancies without known placental pathology, and 3) a refined RPC for uncomplicated term pregnancies. These placental clocks are highly accurate estimators of GA based on placental tissue; e.g., predicted GA based on RPC is highly correlated with actual GA (r>0.95 in test data, median error less than one week). We show that epigenetic clocks derived from cord blood or other tissues do not accurately estimate GA in placental samples. While fundamentally different from Horvath's pan-tissue epigenetic clock, placental clocks closely track fetal age during development and may have interesting applications.

Unbiased data analytic strategies to improve biomarker discovery in precision medicine

Omics technologies promised improved biomarker discovery for precision medicine. The foremost problem of discovered biomarkers is irreproducibility between patient cohorts. From a data analytics perspective, the main reason for these failures is bias in statistical approaches and overfitting resulting from batch effects and confounding factors. The keys to reproducible biomarker discovery are: proper study design, unbiased data preprocessing and quality control analyses, and a knowledgeable application of statistics and machine learning algorithms. In this review, we discuss study design and analysis considerations and suggest standards from an expert point-of-view to promote unbiased decision-making in biomarker discovery in precision medicine.

No evidence for association of MTHFR 677C>T and 1298A>C variants with placental DNA methylation

Background

5,10-Methylenetetrahydrofolate reductase (MTHFR) is a key enzyme in one-carbon metabolism that ensures the availability of methyl groups for methylation reactions. Two single-nucleotide polymorphisms (SNPs) in the MTHFR gene, 677C>T and 1298A>C, result in a thermolabile enzyme with reduced function. These variants, in both the maternal and/or fetal genes, have been associated with pregnancy complications including miscarriage, neural tube defects (NTDs), and preeclampsia (PE), perhaps due to altered capacity for DNA methylation (DNAm). In this study, we assessed the association between MTHFR 677TT and 1298CC genotypes and risk of NTDs, PE, or normotensive intrauterine growth restriction (nIUGR). Additionally, we assessed whether these high-risk genotypes are associated with altered DNAm in the placenta.

Results

In 303 placentas screened for this study, we observed no significant association between the occurrence of NTDs (N = 55), PE (early-onset: N = 28, late-onset: N = 20), or nIUGR (N = 21) and placental (fetal) MTHFR 677TT or 1298CC genotypes compared to healthy pregnancies (N = 179), though a trend of increased 677TT genotype in PE/IUGR together was observed (OR 2.53, p = 0.048). DNAm was profiled in 10 high-risk 677 (677TT + 1298AA), 10 high-risk 1298 (677CC + 1298CC), and 10 reference (677CC + 1298AA) genotype placentas. Linear modeling identified no significantly differentially methylated sites between high-risk 677 or 1298 and reference placentas at a false discovery rate < 0.05 and Δβ ≥ 0.05 using the Illumina Infinium HumanMethylation450 BeadChip. Using a differentially methylated region analysis or separating cytosine-guanine dinucleotides (CpGs) by CpG density to reduce multiple comparisons also did not identify differential methylation. Additionally, there was no consistent evidence for altered methylation of repetitive DNA between high-risk and reference placentas.

Conclusions

We conclude that large-scale, genome-wide disruption in DNAm does not occur in placentas with the high-risk MTHFR 677TT or 1298CC genotypes. Furthermore, there was no evidence for an association of the 1298CC genotype and only a tendency to higher 677TT in pregnancy complications of PE/IUGR. This may be due to small sample sizes or folate repletion in our Canadian population attenuating effects of the high-risk MTHFR variants. However, given our results and the conflicting results in the literature, investigations into alternative mechanisms that may explain the link between MTHFR variants and pregnancy complications, or in populations at risk of folate deficiencies, are warranted.

Electronic supplementary material

The online version of this article (10.1186/s13148-018-0468-1) contains supplementary material, which is available to authorized users.