If you sometimes wonder about the quality of data coming from today’s popular affinity-based DNA methylation analysis methods, you‘re not alone. In fact, researchers at Garvan Institute of Medical Research in Sydney Australia decided to do something about it by studying the techniques and learning that methylated DNA immunoprecipitation (MeDIP) and methyl-CpG binding domain-based capture (MBDCap) approaches are often lugging around more baggage than airport security.
That baggage shows up in the form of bias and similarly to air travel lately, it can cost you. Each of the methods has its own flaws, but the types of effects fell into three categories: CpG density, amplification bias, and copy number variation.
The ambitious Aussies, led by Mark Robinson, compared data from MeDIP and MBDCap experiments and found some interesting results:
- MeDIP and MBDCap each grab different genomic regions based on CpG density. MBDCap gave higher overall enrichment, especially in CpG-rich fragments.
- Whole-genome amplification steps can introduce bias by under-representing GC-rich regions; exactly where you’d want to look for DNA methylation.
- Differentially methylated regions (DMRs) were more easily seen on sequencing platforms in areas of high enrichment, like CpG-rich regions, as they tend to get better sequence coverage, while DMRs in less CpG-dense regions are easier to spot using microarrays.
- Copy number variations (CNVs) can make tracking epigenetic changes a headache if you don’t account for it, especially in methylomes riddled with CNVs, like cancers. The same thing is true in ChIP experiments as well.
The good news is that despite the baggage that comes along with DNA methylation affinity protocols, much of the bias effects that were observed can be taken care of with just a little extra consideration. As the saying goes: Knowledge is power!
Take a look at the full article to make sure your bias baggage is all accounted for at Genome Research, November 2010.