Tim Gardner, of Boston University, gave an interesting talk this morning on the "network biology" approaches he's using in his lab. Interesting in two distinct ways.
One: interesting in itself. Looking for networks by comparing the RNA expression of cells in different conditions and looking for correlations (or, more precisely, according to Gardner, "mutual information") between the expression of regulatory sequences and genes is a neat way of learning more about how cells actually work, a subject on which we are often remarkably ignorant. Key factoid (if I understood it correctly): in E. coli, the best studied bacterium, researchers currently don't have any idea of how three quarters of the genes are regulated. And that shouldn't be taken as meaning that we understand fully how the other quarter is regulated -- just that in those cases we have some leads on the subject (and, to be fair, in some of those cases much more than that).
Second interesting thing: Gardner's take-home message is the exact opposite of the view taken by Drew Endy and his colleagues. Gardner argues that because we have very few well characterised "components" with which to build entirely novel mechanisms and don't really understand how to do so we should concentrate on learning how natural cells work through building network models and get our miracles by tweaking these natural systems. Endy's position is that working out how natural systems actually work is extraordinarily hard (remember that ignorance over three quarters of E. coli) and that we should instead try and build simple things which we do understand. In this respect synthetic biology exists as a counterpoint, or alternative, to systems biology, network biology and other attempts to uncover the ways life actually works.
In part, this is the division between science and engineering. Endy and many of his colleagues at MIT are engineers, and they think in terms of designing well characterised systems, not of understanding very poorly characterised systems such as those that four billion years of evolution have left littering the face of the earth. As Endy puts it, if you were faced with a very complex, very buggy, awesomely antique software system which had been re-worked billions of times, with no notes at all to reveal what all that rewriting was meant to accomplish, or any really well understood sense of what its operating principles were, wouldn't you rather design something new from scratch?
The idea that synthetic biology offers that ability to do wholly new things is often seen as underwriting its practical or commercial possibilities. But it is also, at a more fundamental level, an epistemological distinction that sets this new discipline apart from its predecessors, offering real intellectual novelty. If, that is, Endy and his engineering colleagues can really deliver. Otherwise, it's biology as usual -- even if that biology is, as Garnder's talk was, very interesting in its own right.
I'll try and get a sense of which side of this debate the people attending the conference can be found on; if I turn up anything, I'll report back.
Update: I originally characterised the "mutual information" approach as a way of looking at things "more loosely but more productively", but Rob put me right.