Localized DNA Computation

Abstract

Recently, solution-based systems for DNA computation have demonstrated the enormous potential of DNA nanosystems to do computation at the molecular-scale. These use DNA strands to encode values and use DNA hybridization reactions to perform computations. But most of these prior DNA computation systems relied on the diffusion of DNA strands to transport values during computations. During diffusion, DNA molecules randomly collide and interact in a three-dimensional fluidic space. At low concentrations and temperatures, diffusion can be quite slow and could impede the kinetics of these systems whereas at higher concentrations and temperature, unintended spurious interactions during diffusion can hinder the computations. Hence, increasing the concentration of DNA strands to speed up DNA hybridization reactions has the unfortunate side effect of increasing leaks, which are undesired hybridization reactions in the absence of input strands. Also, diffusion-based systems possess global states encoded via concentration of various species and hence exhibit only limited parallel ability.To address these challenges, this dissertation describes a novel design for DNA computation called a localized hybridization network, where diffusion of DNA strands does not occur. Instead all of the DNA strands are localized by attaching them to an addressable substrate such as DNA nanotrack and DNA origami. This localization increases the relative concentration of the reacting DNA strands thereby speeding up the kinetics. This dissertation demonstrated a localized hybridization network that executed a chain reaction of five DNA hybridizations which executes faster than non-localized DNA reactions.Another advantage of this approach is that each copy of the localized hybridization network operates independently of each other, allowing for a high level of parallelism. Localized hybridization networks also allow one to reuse the same DNA sequence to perform different actions at distinct location on the addressable substrate, increasing the scalability of such systems by exploiting the limited sequence space. An advantage of localized hybridization computational circuit is sharper switching behavior as information is encoded over the state of a single molecule. This also eliminates the need for thresholding as computation is performed locally eliminating the need for a global consensus. There are many applications for localized hybridization networks. These include counting the number of disease marker molecules in a patient, detecting various cancer DNA sequences, and detecting and distinguishing bacteria by their distinguishing DNA. The results from localized DNA hybridization reactions may also be of practical use in performing surface computation on cellular membranes for disease detection and prevention.