What is the benefit of using NGS vs microarray technique for DNA methylation? Seems current methylation arrays, e.g. infinium humanmethylation450 used by TCGA, already provided very high density of coverage, although not down to each single nucleotide, but good enough to discovery methylation pattern along the genome. So is there any reason we have to use NGS rather than microarray for DNA methylation studies?
I disagree with Matthias. As Xin has correctly said, the density of the HumanMethylation450 array is such that it covers methylatioh to a good degree, as well as other arrays such as the Nimblegen tiling arrays.
Also, NGS Bisulphite sequencing is just too expensive to be practically used at the moment at a decent read depth, even reduced representation bisulphite sequencing is quite a financial burden vs. a methylation array.
I totally agree that array can only test a small percentage of pre-selected regions of the genome, but for most if not all scenarios, the major interest of DNA methylation studies is focused on how it will affect expression of genes. So for the ~20,000 human genes, we have on average ~22 probes for each gene, is that still too far from an good coverage?
It's a balance between practicality and comprehensiveness. NGS will certainly give you a more complete picture of the methylome (at least for the 5mC portion), but it is at least an order of magnitude more expensive and will require substantially more analysis (which is to be expected of any discovery-based project). The array was built with the intention of covering as many methylation sites that have at least some evidence of being interesting/relevant. The sites are heavily biased to coding and promoter regions (although there are some throughout the genome). The 450k array remains quite popular (in a time of decreasing array popularity) because it fits what many methylation researchers need. NGS methods are also popular, but remain out of reach for many due to cost and analysis issues. Unless you're in pure discovery mode, most don't seem to want a lot of data they can't interpret. As the methylome becomes better understood over time, the shift from arrays to NGS will likely accelerate.
The price differences between the arrays currently available and the NGS methods are quite high. This becomes especially important when a large amount of samples are to be analysed but I would say that the array could be a filtering step for the future analysis of more interesting samples (less number) using NGS methods; do you agree?
Not really! Since I wrote this post Agilent have come out with their methylSeq product which is probably better than the 450k's, and other NGS technologies have come down in price.
- Badges
- Science method