Friday, February 04, 2005

Genomics the hard way

(UPDATED 8 Feb). In today's Proceedings of the National Academy of Sciences, the Shewanella Federation, a consortium of genomics-oriented labs and companies, will describe how they have "fleshed out" the genome of S. oneidensis, a bug which is capable of sequestering heavy metals and might be useful for bioremediation.

Sequencing of long stretches of DNA has become so fast, these days, that publishing a newly sequenced microbial genome hardly raises an eyebrow anymore. The problem researchers now face is that the work has only begun with the raw sequence. Much work goes into annotating the sequence, that is, figuring out which regions code for proteins, and how those proteins contribute to the life cycle of the microbe. One can take a shortcut if sequence analysis turns up a sequence for which the function in a close relative is known. But genomic data also yield large numbers of so-called "hypothetical genes," consisting of sequences which appear to be correctly configured to specify a protein, but which cannot as yet be confirmed as actually making something, let alone for what function.

The consortium's work here was huge: nearly 40% of the potential genes in S. oneidensis, nearly 1500 sequences, were considered hypothetical. They decided to tackle the whole set of hypotheticals by analysing large numbers of bacterial mRNAs and proteins using high-throughput methods. They managed to positively identify about a third of the hypothetical genes, about 540, as actually being made. But a larger number, 592, were identified using the protein efforts alone, and in fact some of the "dropout" candidates (identified by protein, but not protein+ mRNA) could be reconsidered for technical reasons. So they think that the protein test was probably the more critical for deciding if a given stretch of DNA was actually a gene.

I think the interesting trends here are the massive collaboration going on, and the careful selection of target organisms. Functional genomics is expensive to do, and these projects end up looking like big science as more commonly seen in physics. The worry is that these mammoth projects will suck the funds away from smaller scale stuff. I'm a small-scale guy myself, but I am really fascinated by the possibilities here.

But you don't want to do it on joe microbe. To throw that much money at a problem you'd better want to know answers pretty badly.

No comments: