With everyone focused on their own daily grind, its easy to gloss over some of life’s more philosophical questions like: “Does a falling tree make a sound if no one is there to hear it?” “What is the sound of one hand clapping?” or even “Are my methylome studies really accurate?” In an Epigenomics review article, Australian scientists Mark Robinson, Susan Clark and cohorts took the time to ponder at least one of those questions (hint: it’s the methylome one) and concluded that we should be taking mathylome data with a grain of salt.
Digging Through The Methylome Toolbox
There are a bunch of techniques to study DNA Methylation but this article focused on whole genome, second generation sequencing based ones, which they broke down into three categories, each with their own pros and cons in terms of cost, coverage and resolution:
- Chemical Conversion: These techniques use bisulfite conversion (BS) prior to sequencing to identify methylated cytosines. Examples include: whole-genome shotgun bisulfite sequencing (WGSBS), and reduced representation bislulfite sequencing (RRBS).
- Enzymatic Digestion: Uses restriction enzymes to cleave DNA in a methylation-specific way and learn where methylated sites are located. Some examples: HpaII enrichment by ligation-mediated PCR (HELP-seq) and methylation sensitive cut counting (MSCC)
- Methylated DNA Enrichment: By using 5-mC antibodies or methyl-binding domain (MBD) proteins researchers can capture methylated DNA fragments and detect methylated regions. Examples are: MeDIP-seq, MIRA-seq and MBD-seq
It’s All About the Methylome Protocol
After comparing the various methods WGSBS turned out to be the best thing going in terms of performance, by having the best coverage and resolution available. It also has the highest cost, taking it off the table for many labs. All the other techniques that were surveyed ended up trading some combination of coverage and resolution to go easier on the wallet, and introducing a certain amount of bias in the process.
- Short-read mapping bias: Sequencing reads are mapped by either position or sequence match. When position is used, like in enrichment or enzymatic methods, you lose single-base resolution, and on the other hand anything using bisulfite causes sequence specificity to suffer with the reduction of bases from four to three. Read count of a fragment can also be affected by GC content and the “mappability” of that region.
- CpG coverage bias: Even though it’s at a lower resolution, enrichment methods have an easier time getting good coverage of CpG sites since they capture methylated regions before sequencing, while WGSBS takes many times more reads to get the same coverage.
- CpG density bias: Enrichment reagents are known to work better in CpG dense regions, causing the data to get a little skewed. There are ways to correct for this though, like comparing results to a fully methylated standard or applying some bioinformatics brawn.
- Copy number bias: Genetic variations, such as aberrant copy number, are common in diseased genomes like cancer. Separating the genome according to methylation using enrichment or enzymatic approaches can cause those variable areas to be misrepresented later on during analysis.
There are three things you’d want in an ideal methylome protocol: low cost, high resolution and great coverage. For now, you only get to choose which two are the most important. Hey, nothing’s perfect right?
The Future Looks Bright
Don’t go tossing out your samples and looking career change just yet though, because new technologies are on the horizon that could revolutionize DNA methylation analysis. Third-generation sequencing platforms are popping up that could potentially allow a direct, single-base resolution view of the methylome.
In the meantime, while it may be tempting to just wait for technology to solve all our problems, there are also some ways to tame the biases in your current techniques.
The first, as the authors point out, is to understand the limitations of your protocol. They urge the research community to conduct systematic platform comparisons so that the benefits of each are fully understood.
Also, there are now several high-resolution reference methylomes available though the International Human Epigenome Consortium (IHEC) that can be used to assess techniques.
Find out which methylome you’re actually studying at Epigenomics, August 2010