Evaluating Reliability of DNA Methylation Measurement

Area of Science:

• Epigenetics
• Genomics
• Biostatistics

Background:

• DNA methylation is a key epigenetic regulator influencing human traits and diseases.
• Microarray platforms such as Illumina Infinium HumanMethylation450 (HM450) and MethylationEPIC (EPIC) are crucial for epigenome-wide association studies (EWASs).
• Assessing technical variability is essential for the reliability of EWAS findings.

Purpose of the Study:

• To introduce and evaluate the intraclass correlation coefficient (ICC) for quantifying technical variation in DNA methylation data at the single-CpG site level.
• To characterize the distribution of ICC values and classify CpG sites based on reliability.
• To provide practical recommendations for interpreting EWAS results in light of technical variability.

Main Methods:

• Utilized intraclass correlation coefficient (ICC) to compare within- and between-replicate methylation level variations.
• Employed a mixture of truncated normal and normal distributions to model ICC distributions.
• Clustered CpG sites into low- and high-reliability groups based on ICC values.

Main Results:

• Demonstrated the utility of ICC in assessing technical variation in methylation array data.
• Characterized ICC distributions and identified distinct groups of CpG sites with varying reliability.
• Provided a framework for identifying and handling CpG sites with low technical reliability.

Conclusions:

• The ICC is a valuable metric for evaluating the technical reliability of CpG sites in methylation array data.
• CpG sites with low ICC values represent higher technical variation and require careful consideration in EWAS.
• Implementing ICC assessment can enhance the robustness and interpretability of epigenome-wide association studies.