How does the Genome Analyzer fit into your research?

The interest of the lab is to understanding the genetic and epigenetic changes that result in phenotypic variation. Our main interest is in plants, particularly Arabidopsis, although we work with human cells as well. Our lab is very much discovery driven, not hypothesis-driven. Our rational is: if we use this technology at the whole genome level, we will surely discover something interesting. That’s about the only hypothesis we have going into such as study and we have never been disappointed by this approach. For example, we know that we don’t understand all the transcripts in the genome. Now we are moving discovery to the sequence level, which is something we would have wanted to do in the first place, but we couldn’t afford to do these experiments at this scale.

So, by necessity, we used array-based methods with inherently lower resolution. Ultimately, the level of resolution that we need to understand gene structure, to understand methylation, to understand DNA sequence variation, is sequence. We are formulating a plan with our colleagues here in the U.S. and around the world to carry out DNA sequencing for over 1,001 Arabidopsis strains. When completed, this sequence data will allow for whole genome association studies with an unprecedented level of resolution. We are extremely excited about this plan, which would not have been conceivable without the Genome Analyzer.

Can you tell us about your methylation research using the Genome Analyzer?

We started sequencing the methylome in March. In a matter of five months we are now very close to competing this work. This was unthinkable a year ago. You need to be in the 40-50X coverage range because you are looking at a population of cells. If you look at 50 chromosomes there may be methylation, or methylation at a particular site, so if you want to get quantification of how much there is at a specific site you need to sequence very deep. So the question is not only is this specific cytosine methylated or not, but also is it methylated 20% of the time, 40%, or 80% of the time. If we were doing Sanger sequencing and doing 50-fold coverage, that would be like sequencing the entire human genome at ~2X coverage.