Improved filtering of DNA methylation microarray data by detection p values and its impact on downstream analyses

Area of Science:

• Genetics
• Bioinformatics
• Epigenetics

Background:

• DNA methylation microarrays are widely used for epigenome-wide association studies (EWAS).
• Conventional filtering methods for probe detection p-values are insufficient, leading to spurious methylation calls, particularly for Y-chromosome probes in females.
• This compromises data quality and hinders replication in EWAS.

Purpose of the Study:

• To develop and evaluate an improved method for calculating detection p-values using non-specific background fluorescence.
• To assess the effectiveness of the proposed filtering approach compared to conventional methods in reducing spurious values and improving EWAS reliability.
• To provide guidance for preprocessing DNA methylation microarray data for both 450K and EPIC platforms.

Main Methods:

• Developed an alternative approach to calculate detection p-values by incorporating non-specific background fluorescence.
• Evaluated the new method by analyzing Y-chromosome probe detection in 2755 samples across 17 studies using the 450K microarray.
• Assessed the masking of large outliers between technical replicates and their downstream impact via EWAS reanalysis.

Main Results:

• The proposed method accurately marks most Y-chromosome probes in females as undetected, unlike conventional approaches.
• It removes a median of only 0.14% of data per sample while effectively identifying more large outliers (30% vs. 6%) between technical replicates.
• The improved filtering identified strong DNA methylation associations with chronological age previously obscured by outliers in a large EWAS.

Conclusions:

• The developed method offers a more accurate way to filter DNA methylation microarray data, essential for robust EWAS.
• Guidance is provided for filtering both 450K and EPIC microarray data as a critical preprocessing step.
• An implementation is available in the ewastools R package to facilitate comprehensive quality control for DNA methylation microarrays.