9 December 2005
Drew Endy
Division of Biological Engineering
MIT
Nothing in biology's past makes sense except in the light of evolution. Going forward, we can either engineer disposable biological systems (that are not designed to evolve) or learn how to program reproducing biological machines that we still somehow understand. I'll discuss a simple framework for mapping the architecture and function of reproducing machines across three different time-scales of physical performance, and the idea that the observed architecture of reproducing machines might reflect a compromise between sub-system modularity and the detection of errors during machine replication---at this point, I'll argue that evolution may select for a form of "anti-modularity" and introduce some experiments that we are attempting in yeast in order to begin to test this idea.