Navigating Illumina DNA methylation data: biology versus technical artefacts.

Area of Science:

• Epigenetics and Genomics
• Molecular Biology
• Bioinformatics

Background:

• Illumina BeadChip arrays are vital for genome-wide DNA methylation profiling in diagnostics.
• Comprehensive quality assessment is difficult with diverse tissues and preparation methods.
• Suboptimal samples pose challenges in distinguishing biological signals from technical noise.

Purpose of the Study:

• To develop novel quality control metrics for DNA methylation data, especially from suboptimal samples.
• To assess the impact of sample ranking on Illumina-like normalization algorithms.
• To enhance the reproducibility and reliability of DNA methylation profiling.

Main Methods:

• Introduction of three novel quality control scores: DB-Score, BIN-Score, and CM-Score, based on biological methylation features.
• Benchmarking of these scores across independent cohorts and varying sample types.
• Investigation of sample ranking effects on Illumina-like normalization and validation with whole-genome bisulfite sequencing data.

Main Results:

• The proposed quality control scores effectively differentiate biological effects from technical artifacts in challenging samples.
• Sample ranking significantly impacts beta values in Illumina-like normalization, affecting downstream differential methylation analysis.
• Consistent use of a pre-defined normalization sample is crucial for robust and reproducible results.

Conclusions:

• The developed quality control scores offer a robust method for assessing DNA methylation data quality.
• Standardizing the normalization process by fixing the ranking sample is essential for improving the reproducibility of genome-wide methylation studies.
• The study provides practical recommendations and R functions to enhance DNA methylation analysis quality assurance.