Computing Comes to Life
Biochemical Circuits and Networks
The analogy between metabolic regulators and digital logic was already noticed 40 years ago. In 1961 Monod and Jacob wrote about genetic circuits and switching networks, and they described how activator and repressor proteins could be organized into systems that would function as memory elements and oscillators. Other authors soon explored the connection between molecular biology and digital computing in greater depth and detail; indeed, for several years the theme was a frequent one in the Journal of Theoretical Biology and the Bulletin of Mathematical Biophysics.
The main focus of these early studies was on using digital models as a way of understanding events in the living cell. The Boolean approximation was a way of avoiding an unwieldy analysis of a complex chemical web. To follow all those molecular interactions in complete detail would have required tracking the concentrations of innumerable molecular species, measuring the rates of chemical reactions, and solving hundreds of coupled differential equations. Pretending that every gene is either on or off reduced the problem to a simpler digital abstraction.
The biological computer turns this idea upside down. Instead of constructing a computational model of biochemistry, you exploit quasi-Boolean biochemistry to do computing. This notion also has a history. In the 1970s Otto Rössler analyzed various coupled systems of chemical reactions that could implement the abstract computers called finite automata. More recently, other groups have looked at schemes of computing based on the catalytic activities of enzymes.
The most novel plan for biologically inspired computing was conceived by Leonard M. Adleman of the University of Southern California. His basic idea is to use the complementary base-pairing of DNA as a pattern-matching engine. Adleman himself and others have demonstrated the feasibility of this idea in experiments where vials of DNA carry out computational tasks in number theory and combinatorics.