Models and Algorithms for Molecular Programming

点击翻译按钮获取中文摘要

Bio-molecules do amazing things in the cell. Some store information, others translate that information to produce "actor" molecules that sense their environment and respond accordingly, via sophisticated networks of reactions. Bio-molecules self-assemble into scaffolds, or form compartments through which select molecules enter and exit. Molecular interaction is also the basis for movement and transport of cargo in the cell. Researchers in the field of DNA computing and molecular programming ask: how can we use nature's versatile bio-molecules in new ways: to compute, or to build structures and molecular robots? My long-term goal is to devise new ways to program molecules, by designing and analyzing algorithms expressed as Chemical Reaction Networks (CRNs) and other molecular programming models, by developing effective schemes to compile such algorithms down to DNA implementations, and by understanding the limits of what is possible. My short-term objectives will focus on two steps in the compilation process. (1) DNA strand displacement systems (DSDs) provide a valuable intermediate description of molecular computations, between CRNs and a DNA implementation. Currently, CRN-to-DSD compilation schemes vary across research labs, with respect to vulnerability to errors or the ease with which they can be implemented or verified. I will study new ways to compile CRNs down to DSDs, both in solution and on surfaces, so as to simplify the multi-stranded complexes involved, thereby reducing errors and using strands efficiently. I also want to understand limits to efficient compilation schemes. (2) DSDs are further compiled down to DNA strands, by associating a sequence of A,C,G,T bases with each abstract DSD domain. Kinetic simulators, such as Caltech's Multistrand simulator, are used to ensure that DNA implementations faithfully replicate the intended DSD behaviours. We will improve the accuracy and usability of Multistrand, by inferring improved kinetic parameters and by developing coarse-grained approximations of Multistrand's stochastic model of DNA reactions. These objectives provide fertile opportunities for collaborative, interdisciplinary, HQP training. A newly recruited postdoc and two graduate students are already working on our objectives, bringing valuable expertise of their own. I will provide a stimulating, supportive environment for these and future advisees, positioning them for meaningful careers. Computations happen in silico, in the brain, and in the cell through chemical reactions. The last of these is perhaps the least understood, and is particularly fascinating because of the ways that tiny bio-molecules both encode the script and are the actors in their environments. Our research will advance our ability to program bio-molecules in creative ways, and hopefully pave the way for exciting new applications of computing principles in synthetic biology.