Controlling for conservation in genome-wide DNA methylation studies

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• High-throughput DNA methylation studies commonly compare methylation levels across functional regions.
• Previous analyses suggested higher methylation in coding regions versus introns or UTRs, implying novel epigenetic regulation.

Purpose of the Study:

• To identify and characterize a bias in DNA methylation comparisons across region types.
• To develop and evaluate methods for correcting this bias to enable accurate analysis of DNA methylation.

Main Methods:

• Identified a bias in DNA methylation comparisons related to conservation rates, not methylation rates.
• Developed two correction methods: an inference-based matrix completion algorithm and an averaging approach.
• Re-evaluated DNA methylation rates at intron-exon junctions and coding start sites after controlling for conservation.

Main Results:

• A significant bias was found in published DNA methylation data, leading to false detection of differences across region types.
• Controlling for conservation at coding start sites eliminated observed differences in DNA methylation rates.
• The magnitude of previously reported differences in methylation at intron-exon junctions was found to be greatly exaggerated.

Conclusions:

• A bias originating from conservation rate differences falsely indicates variations in DNA methylation intensity and extent.
• Characterized the bias and its implications, providing methods to control for it.
• Enabled accurate study of DNA methylation across functional genomic regions.