Physicist Amok

I attended a New Vistas lecture at the New York Academy of Sciences this evening.  David Botstein of Princeton’s Sigler Institute hosted Mike Elowitz, formerly of Rockefeller, now at Caltech, and Saaed Tavazole, also at the Sigler Institute.

Elowitz studies stochastic variation in clonal populations.  It was a fad for a while, but it was important to break the mindset that identical genes mean identical cells which ruled the minds of most microbiologists.  Circuitously, it led to my project.  He just rehashed the fact that there was cell to cell variation.

Tavazole is looking for gene expression signals that might correlate with ecological niches in bacteria.  However, he’s looking in E. coli K12 MG1655, the standard lab strain that has been passaged in a test tube for forty years.  He found a correlation between the regulation of genes induced by increasing oxygen and decreasing temperature (corresponding to leaving the mammalian gastrointestinal tract for the soil), and then tried to break the connection between the two by cycling between high oxygen, high temperature and low oxygen, low temperature conditions in a chemostat.  Quickly strains evolved which outcompeted the ancestors in this situation, but it is predictably an artifact: the growth rate of the evolved strains goes up very quickly when they are first switched to high oxygen, high temperature, but it quickly decays back to that of the ancestral strain.  The strain has sped up its adaptation, and possibly done something strange like undergoing high speed, reductive cell division when it is shifted between conditions.  He showed that the correlations from the ancestral strain between oxygen and temperature induced genes are weakened or broken, but I’m not sure how significant that is.  If the bacteria are undergoing some wild, uncontrolled cell division, there is no reason to expect their gene expression profiles to be the same.

Systems biology as far as I can tell is a rehashing of the old cybernetics program on genes, wedded with some bits and pieces of statistical genetics.  The endgame appears to be a directed graph, possibly with multiple kinds of arrows, each corresponding to a term in a (possibly stochastic) differential equation.  I have two particular concerns with this program.  First, you can map systems of differential equations , where is a polynomial of low degree, probably two, unambiguously onto such a directed graph (I don’t offer this as a rigorous theorem, merely an observation), but you cannot map partial differential equations, deterministic or stochastic, without an infinite number of edges.  Second, I feel like the program does not allow you to formulate questions that aren’t fairly trivial.  The directed graph models were inherited from chemistry, where they describe reactions in well mixed solutions.  You can talk about sodium ions binding to and coming off from chloride ions, but it won’t tell you about the crystalline forms that form as you saturate the solution.

In lab meeting this morning, I had another of those moments that remind me I’m not in physics anymore.

We apparently have a fairly new -80°C freezer in BL3 (Biosafety Level 3, for those with benign organisms). It doesn’t take the same rack size as the previous one. The old racks are too deep for it. My lab groaned about how many thousands of dollars it would cost to get new, properly sized racks.

I say, “Why don’t we just take a hacksaw to them? Or I know the high energy physics machine shop has a metal bandsaw.� You would have thought I had grown a second head.

It was almost as bad as when I recommended condoms to maintain a partial seal on flasks in the autoclave. They’re the sturdiest membrane for your money, and I’m sure some of the tougher materials will happily survive autoclaving even if straight latex won’t.