Shannon's measure of information is useful for characterizing the DNA and RNA patterns that define genetic control systems. Thus we may measure the amount of pattern which a ribosome has available to it in the mRNA. In E. coli this is about 11.0 bits per site, where one bit is the choice between two equally likely possibilities. A single measurement is meaningless unless we have something to compare it to. Fortunately, we can calculate the information required to select ribosome binding sites from other mRNA sequences. This is 10.6 bits per site, almost identical to the amount available. Thus the amount of pattern at ribosome binding sites is just about the amount of information needed to find the sites. This is a """"""""working hypothesis"""""""": exceptions will either destroy the hypothesis or reveal new phenomena. Several genetic systems now fit the hypothesis, but a few provide striking exceptions which we are actively studying. In particular, the sequences at which bacteriophage T7 bind have twice the information required to locate them. The most likely explanation is that two proteins bind the DNA. The project has three major components: theory, computer analysis and genetic engineering experiments. Computer analysis and theory predict that the T7 promoters and RepA binding sites have anomalies, so we are performing experiments to find out why. My theoretical work can be divided into several levels. Level 0 is the study of genetic sequences bound by proteins or other macromolecules, briefly described above. The success of this theory suggested that other work of Shannon should also apply to molecular biology. Level 1 theory introduces the more general concept of the molecular machine, and the concept of a machine capacity equivalent to Shannon's channel capacity. In Level 2, the Second Law of Thermodynamics is connected to the capacity theorem, and the limits on the functioning of Maxwell's Demon become clear. Publications were completed at all three levels.
