Related Experiment Video
Updated: Sep 25, 2025
14:56
Sample Preparation to Bioinformatics Analysis of DNA Methylation: Association Strategy for Obesity and Related Trait Studies
Published on: May 6, 2022
4.7K
Batch-effect detection, correction and characterisation in Illumina HumanMethylation450 and MethylationEPIC BeadChip
Jason P Ross1, Susan van Dijk2, Melinda Phang3
1Human Health Program, Health and Biosecurity, CSIRO, Sydney, Australia. jason.ross@csiro.au.
Clinical Epigenetics
|April 30, 2022
Summary
Batch effects persist in Illumina methylation data despite preprocessing, impacting epigenome-wide association studies (EWAS). Careful application of batch-effect removal tools and post hoc diagnostics are crucial for reliable EWAS results.
Area of Science:
- Genomics
- Epigenetics
- Bioinformatics
Background:
- Genomic technologies are susceptible to batch effects, potentially reducing experimental power and yielding false positives in epigenome-wide association studies (EWAS).
- The Illumina Infinium Methylation BeadChip is widely used for EWAS, but batch effect characteristics remain poorly understood.
- Controlling for batch effects is critical given the subtle biological phenotypes often investigated in EWAS.
Purpose of the Study:
- To investigate the nature and impact of batch effects on Illumina Infinium Methylation BeadChip data.
- To identify persistent batch effects and probes susceptible to erroneous correction.
- To provide practical workflows and reference data for managing batch effects in EWAS.
Main Methods:
- Applied ComBat and Harman batch-effect removal software to two in-house and three large public datasets (1214 HumanMethylation450 and 1094 MethylationEPIC BeadChips).
- Analyzed residual batch effects associated with processing day, slide position, and array position.
- Reviewed literature to identify instances where batch-effect prone probes were reported as significant in EWAS.
Main Results:
- Despite preprocessing, residual batch effects were detected and associated with processing variables.
- Consistently identified 4649 probes requiring significant batch correction across all datasets.
- Discovered a set of probes that were erroneously corrected by batch-effect removal methods.
- Provided reference matrices for batch-effect prone and erroneously corrected features across diverse populations and biosamples (blood, buccal, saliva).
Conclusions:
- Batch effects are inherent in high-quality genomic data, necessitating a proactive strategy in experimental design for EWAS.
- Batch-effect removal tools are valuable for reducing technical variance in Infinium Methylation data but require careful application and post hoc validation.
