Genome biology has an update on metagenomics, the genomic sequencing of multiple related organisms all in one go. Metagenomics was first invented for the purpose of getting sequence data from microbes which cannot be grown in the lab. (This is the vast bulk of all microbes). Places like acid pits in mines, the open ocean, or even the air in Manhattan are brimming with life, but scientists are not able to grow the individual organisms yet to the point where they could get enough DNA for sequencing.
Metagenomics gets around this problem by big improvements on the front and back ends of the sequencing effort. On the front end, DNA is just taken and sheared "shotgun" from an environmental source, and clever methods are used to clone the DNA (chemically join it to known sequences to make a recombined DNA package which bacteria will copy). This means that the DNA is recovered in a fairly unbiased way.
The sequencing is done all at once on the mixture of all DNAs which were available in the environmental sample.
On the back end, computer analysis is used to group (assemble) sequences, without knowing the source organism. In many cases, the sequences are similar enough that source organisms can be identified. The frequency with which a particular sequence is obtained hints at the relative numbers of its host organism in the environment.
What's interesting to me about the analysis is that the results produce a portrait of an ecosystem. Comparing the sequences found in one environment to another set can highlight the specializations characteristic of the whole assemblage of organisms, allowing meta-comparisons analagous to "Hollywood is all about movies and New York is about finance." (Probably at a similar level of resolution). For example, the open ocean (Sargasso Sea) sequences showed a huge variety and frequency of rhodopsin-like proteins, suggesting that photosynthesis is probably a property of very many pelagic species. Again, some have not even been identified yet except for their sequence! Soil samples show a diversity of cellubiose phosphorylase genes, again consistent with a community organized around decaying plant matter.
The point is that these descriptions, encompassing a kind of weighted average of the actual inhabitants of a particular spot, portray how microbes actually live in the wild. Microbes always occur in communities, and they rely on each other extensively. As a concrete example, the paper mentions an future effort at metagenomic analysis of the 10,000,000,000-strong community of microbes in and on the human body. There is as much DNA in those microbes as within the human genome, and its detailed makeup changes with every meal, even with every shower.