>

Summary

In this paper I have defined molecular machines and constructed a mathematical model for them that fits many examples in modern molecular biology. The mathematical description of molecular machine operations uses the methods of information theory, for which the hallmark and yardstick is the bit. According to this theory if a molecular machine is exposed to white Gaussian noise, then it should not be possible for it to gain more information than that given by the capacity formula, equation (38), although it may be able to approach this limit.

A theorem, originally proven by Shannon, shows that molecular machines can act precisely despite the ubiquitous presence of thermal noise. This is not a quantum nor chemical-bonding effect, but rather it arises from the degree of complexity that a molecular machine can attain by evolving a molecular coding scheme. The channel capacity should be a useful criterion for understanding and designing molecular machines [Maniatis et al., 1982,Beaucage & Caruthers, 1981,,Pabo, 1983,Rastetter, 1983,Wetzel, 1986,Lesk, 1988].

I thank Herb Schneider, John Spouge, Randy Smith, and Peter Basser for many fun and useful discussions; Doris Schneider, Sislin Schneider, Sue Aldor and Maureen Manns for their support; and Stephen Altschul, Steve Garavelli, Rob Harrison, Jim Hofrichter, Andrzej Konopka, Peter Lemkin, Sarah Lesher, David Lipman, Joe Mack, Jake Maizel, Hugo Martinez, Ranjan Muttiah, Howard Nash, Peter Rogan, Denise Rubens, Morton Schultz, Bruce Shapiro, and R. Michael Stephens for critically reading the manuscript. I also thank Larry Gold for supporting the preliminary stages of this project under NIH grant GM28755, Ron Fox for pointing out the form of f_D(r), Richard Pastor for pointing out the need to account for conformational substates, and Gray Abbott for useful discussions on Fourier analysis. The gumball machine was manufactured by Superior Toy & Mfg. Co. Inc., Chicago, IL. I thank Charles Rockwell for the photography of Fig. 1 .