Patterns of Reliability: Assessing the Reproducibility and Integrity of DNA Methylation Measurement

Abstract

SUMMARY DNA methylation plays an important role in both normal human development and risk of disease. The most utilized method of assessing DNA methylation uses BeadChips, generating an epigenome-wide “snapshot” of >450,000 observations (probe measurements) per assay. However, the reliability of each of these measurements is not equal, and little consideration is paid to consequences for research. We correlated repeat measurements of the same DNA samples using the Illumina HumanMethylation450K and the Infinium MethylationEPIC BeadChips in 350 blood DNA samples. Probes that were reliably measured were more heritable and showed consistent associations with environmental exposures, gene expression, and greater cross-tissue concordance. Unreliable probes were less replicable and generated an unknown volume of false negatives. This serves as a lesson for working with DNA methylation data, but the lessons are equally applicable to working with other data: as we advance toward generating increasingly greater volumes of data, failure to document reliability risks harming reproducibility.

A computational solution for bolstering reliability of epigenetic clocks: Implications for clinical trials and longitudinal tracking

“…First, we calculated ICCs for 1,273 individual CpGs that are part of five existing clocks-the Horvath multi-tissue predictor (Horvath1), the Horvath skin-and-blood clock (Horvath2), the Hannum blood clock (Hannum), the Levine DNAmPhenoAge clock (PhenoAge), or the Lu telomere length predictor (DNAmTL) (Table S1) (Horvath 2013;Hannum et al 2013;Levine et al 2018;Lu, Seeboth, et al 2019 S2). Low-reliability clock CpGs tend to have more extreme values (near 0 or 1) and lower variance (Figures 1B-C), consistent with prior genome-wide analyses (Sugden et al 2020;Bose et al 2014). CpGs with strong associations with mortality (hazard ratios after adjusting for age and sex) or with chronological age tended to have higher ICCs (Figure 1D).…”

“…Alas, previous studies have shown that the majority of individual CpGs are unreliable, yielding surprisingly variable methylation values when the same biological specimens are measured multiple times due to technical variance inherent to the array (Sugden et al 2020;Logue et al 2017;Bose et al 2014). In many cases, technical variance exceeds the biological variance for the DNAm levels at individual CpGs, as quantified by the intraclass correlation coefficient (ICC) metric.…”

Rare genetic variation at transcription factor binding sites modulates local DNA methylation profiles

“…While previous studies have shown that epigenetic changes due to regulatory mutations often lead to regional changes affecting clusters of multiple CpGs, it should be noted that for the majority of SNV-TFBS in our study (74%), we identified only a single associated outlier CpG. There are several possible explanations for this observation: (i) the EPIC array that we used to interrogate DNA methylation only has relatively sparse genomic coverage, sampling only a small subset of CpGs at most genomic loci; (ii) due to the large number of SNV-TFBS we interrogated, some of the pairwise associations with outlier methylation occurred by chance, and (iii) differential methylation of single CpGs might sometimes result from inaccurate measurements due to poorly performing probes or other technical artefacts, and does not represent true epigenetic variation [ 39 ]. To minimize this latter possibility, we performed pre-processing of the methylation dataset including the removal of potentially confounded probes and normalization.…”

Applying epigenetic science to the understanding of eating disorders: a promising paradigm for research and practice

“…Of note, to date, all epigenome-wide methylation studies in eating disorders have addressed anorexia nervosa, meaning that studies on bulimia nervosa, binge eating disorder and related eating disorders are needed. From a methodological standpoint, further studies into the reliability and functional relevance of identified DNA methylation marks are warranted [ 35 ]. Finally, it is well established that methylation levels of CpGs in certain genes are allele/genotype-dependent [ 36 – 38 ].…”