DAN-based string rewrite computational systems

We describe a DNA computing system called programmed mutagenesis. prove that it is universal, and present experimental results from a prototype computation. DNA is a material with important characteristics, such as possessing all the information necessary for self-reproduction in the presence of appropriate enzymes and components, simple natural evolution mechanism, and miniature scale, all of which make it an attractive substrate for computation. For computer science, using single DNA molecules to represent the state of a computation holds the promise of a new paradigm of composable molecular computing. For biology, the demonstration that DNA sequences could guide their own evolution under computational rules may have implications as we begin to unravel the mysteries of genome encoding. Programmed mutagenesis is a DNA computing system that uses cycles of DNA annealing, ligation, and polymerization to implement programmatic rewriting of DNA sequences. We report that programmed mutagenesis is theoretically universal by showing how Minsky's 4-symbol 7-state Universal Turing Machine can be implemented using a programmed mutagenesis system. Each step of the Universal Turing Machine is implemented by four cycles of programmed mutagenesis, and progress is guaranteed by the use of alternate sense strands for each rewriting cycle. We constructed a unary counter, an example programmed mutagenesis system, and operated it through three cycles of mutagenesis to gather efficiency data. We showed that the counter operates with increasing efficiency, but decreasing overall yield. The measured efficiency of an in vitro programmed mutagenesis system suggests that the segregation of the products of DNA replication into separate compartments would be an efficient way to implement molecular computation. Naturally occurring phenomena such as gene conversion events and RNA editing processes are also discussed as possible manifestations of programmed mutagenesis-like systems. Thesis Supervisor: Professor David K. Gifford Title: Professor of Electrical Engineering and Computer Science