Related Experiment Video
Updated: May 26, 2026
13:47
Enhanced Reduced Representation Bisulfite Sequencing for Assessment of DNA Methylation at Base Pair Resolution
Published on: February 24, 2015
Batch effect correction for genome-wide methylation data with Illumina Infinium platform
Zhifu Sun1, High Seng Chai, Yanhong Wu
1Division of Biomedical Statistics and Informatics, Department of Health Sciences Research, Mayo Clinic College of Medicine, 200 First Street, Rochester, MN 55905, USA.
BMC Medical Genomics
|December 17, 2011
Summary
Batch effects in genome-wide methylation data can skew results. Normalization methods reduce some effects, but Empirical Bayes (EB) correction is crucial for accurate analysis of Illumina Methylation BeadChip data.
Area of Science:
- Genomics
- Epigenetics
- Bioinformatics
Background:
- Genome-wide methylation profiling offers insights into gene regulation and therapeutic targets.
- Illumina Human Methylation BeadChip is a widely used platform for methylation studies.
- Methylation data is prone to technical artifacts, notably batch effects, which require careful handling.
Purpose of the Study:
- To evaluate normalization methods for batch effect removal in methylation data.
- To assess the effectiveness of different normalization techniques and Empirical Bayes (EB) correction.
- To improve downstream analysis accuracy by mitigating technical variations.
Main Methods:
- Comparison of three normalization approaches: quantile normalization at average β value (QNβ), "lumi" package normalization, and A/B signal quantile normalization (ABnorm).
- Evaluation of subsequent Empirical Bayes (EB) batch adjustment.
- Utilized three datasets with varying degrees of batch effects from the HumanMethylation27 platform.
Main Results:
- Normalization reduced batch effects, with "lumi" performing best on minor batch effects.
- Substantial batch effects remained in datasets with obvious variations, necessitating further correction.
- Empirical Bayes (EB) correction effectively removed residual batch effects, significantly increasing CpGs associated with outcomes.
Conclusions:
- Batch effects significantly impact genome-wide methylation data analysis from Infinium Methylation BeadChip.
- Normalization alone is insufficient for complete batch effect removal.
- A combined approach of normalization and Empirical Bayes (EB) correction is recommended for robust methylation data analysis.
