An evaluation of processing methods for HumanMethylation450 BeadChip data.

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• Illumina's HumanMethylation450 arrays are cost-effective for high-throughput DNA methylation analysis.
• Technical artifacts like background fluorescence, dye-bias, probe design bias, and batch effects are common concerns.
• Various processing methods exist, targeting single or multiple biases.

Purpose of the Study:

• To evaluate the effect of combining separate approaches for improved signal processing in DNA methylation data.
• To compare the performance of nine different preprocessing methods on four datasets.
• To identify optimal preprocessing strategies for HumanMethylation450 BeadChip data.

Main Methods:

• Applied and compared nine processing methods (within- and between-array) across four datasets.
• Evaluated methods based on variance reduction in technical replicates.
• Assessed improvement in differential DNA methylation signal detection for biological differences.

Main Results:

• Within-array processing consistently improved differential DNA methylation signal detection.
• Combining within-array procedures generally yielded the best results.
• The between-array method Funnorm showed an advantage when batch effects were substantial.
• RUVm batch correction notably improved reproducibility when batch effects dominated signal variation.

Conclusions:

• The within-array combination of Noob + BMIQ consistently enhanced signal sensitivity.
• Combining Noob + BMIQ with RUVm batch correction outperformed all other methods for differential DNA methylation analysis.
• The choice of data processing method depends on the interplay between signal and noise in the dataset.