Related Experiment Video
Updated: May 16, 2026
13:47
Enhanced Reduced Representation Bisulfite Sequencing for Assessment of DNA Methylation at Base Pair Resolution
Published on: February 24, 2015
A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA
Andrew E Teschendorff1, Francesco Marabita, Matthias Lechner
1Statistical Genomics Group, UCL Cancer Institute, University College London, London WC1E 6BT, UK. a.teschendorff@ucl.ac.uk
Bioinformatics (Oxford, England)
|November 24, 2012
Summary
This study introduces BMIQ, a new normalization method for Illumina 450 k DNA methylation data. BMIQ adjusts type 2 probes to match type 1 probes, improving data quality for epigenome-wide association studies.
Area of Science:
- Genomics
- Epigenetics
- Bioinformatics
Background:
- Illumina Infinium 450 k Beadchip is widely used for Epigenome-Wide Association Studies (EWAS).
- Two probe designs (type 1 and type 2) exhibit different methylation distributions and dynamic ranges, potentially biasing EWAS results.
- A statistical challenge is to normalize data effectively across these probe designs.
Purpose of the Study:
- To develop a novel intra-array normalization strategy for Illumina 450 k DNA methylation data.
- To address the bias introduced by differing type 1 and type 2 probe designs.
- To improve the accuracy and robustness of EWAS analyses.
Main Methods:
- Propose BMIQ (Beta MIxture Quantile dilation), a model-based normalization strategy.
- Utilize a three-state beta-mixture model to classify probe methylation states.
- Apply quantile transformation and methylation-dependent dilation to adjust type 2 probe beta-values.
Main Results:
- BMIQ successfully adjusts type 2 probes to match the distribution of type 1 probes.
- Validated on cell-line, fresh frozen, and paraffin-embedded tumor tissues.
- Demonstrates improved robustness, reduced technical variation, and elimination of type 1 enrichment bias compared to existing methods.
Conclusions:
- BMIQ is an effective preprocessing step for Illumina Infinium 450 k data.
- The method enhances data quality for EWAS by normalizing across probe designs.
- BMIQ is freely available for use in epigenetic studies.
