After being reprimanded justifiably by one of the authors of the recent paper on bee colony collapse disorder (for a blog I posted and then removed) I have come to realize they used a very interesting approach to metagenomics that I have not seen used extensively before.
Their paper can be found here.
Here is the problem they were faced with - how to survey an sample for ALL the microbes present including both viruses and cellular microbes. The challenge to this is that some viruses have RNA genomes and thus if one simply extracts DNA and sequences it one will not sample any of the RNA viruses. One approach to such a challenge would be to isolate RNA and make cDNA and sequence to sample the RNA viruses. And then to separately isolate DNA and then sequence it either directly (using shotgun sequencing) or indirectly by first amplifying genes with PCR.
They chose a different approach - to isolate RNA and make cDNA and then sequence it. In doing this they in fact do get a sample of both RNA viruses AND cellular organisms. For cellular organisms, since most of the RNA in a cell is ribosomal RNA they get a sample of that organisms rRNA which can be used to say what type of organisms are present. Thus in one fell swoop they in essence sample ALL the microbes present in a sample. I had blogged about this and criticized them because they did not explain in the paper or in the press releases all of this logic, but one of the authors set me straight so I deleted my blog. Then I started thinking about it and realized that this seemed to be a relatively novel approach to metagenomics.
Now - I am not sure if this method is quantitative or exactly how robust it is, but it does provide an alternative to rRNA PCR (which has all the biases of PCR) and also provides an alternative to separately sequencing RNA and DNA from a sample. Certainly many have used RNA to cDNA and then sequencing to survey RNA viruses before. But usually they do this in material in which the cellular organisms have been first removed so that one does not get overwhelmed by the RNA from those organisms. But here, they used the power of massively high throughput sequencing and turned this "problem" of getting RNA from cellular organisms on its head and used it to sample RNA viruses and cellular organisms at the same time.
I do not know if this has been done before --- maybe other out there know of examples. But whether or not it has been done before, it is an important approach that should be considered in metagenomic surveys and it also suggests that ANYONE doing metagenomic surveys of microbes might want to purify RNA and save it form samples even if one is going to first focus on DNA.