59 / 79
Engineering Approaches to Biomolecular Motors: From in vitro to in vivo Poster Abstracts
54
31-POS
Board 31
Computase: Computing with a Self-assembling 3D Lattice of De Novo Designed
Biomolecular Machines
Kathy Y. Wei
, David Baker.
University of Washington, Seattle, WA, USA.
Although silicon-based computers continue to improve, alternative models of computation have
the potential to supply more energy-efficient and parallelized solutions to domains difficult for
sequential instruction-based semi-conductor computers. While traditional computers require
energy to hold the state of each transistor, proteins will only require energy to change state. We
propose a protein-based computer that self assembles into a 3D cubic lattice of nodes. Each node
is a computationally designed 3-input/3-output protein logic gate “wired” to its 6 lattice
neighbors. The connectedness allows computing processes to be parallel - throughout the volume
of the assembled lattice - and be scalable. For example, the proposed protein computer could
process an image by taking one pixel per node along an entire face of the lattice and thus process
an entire image simultaneously, as opposed to sequentially on traditional computers. Physically,
each node is composed of two components: a scaffold to support self-assembly, and a molecular
machine that integrates and transfers information between lattice neighbors. We have
demonstrated the ability to build multicomponent self-assembling nanostructures and 2D protein
arrays, and to design de novo proteins with atomic accuracy. As a first step, we will design
bistable protein switches that can 1) toggle between two states, and 2) toggle in such a way as to
influence the states of neighboring nodes. The protein design goals of 3D self-assembly, bistable
switches, and directed protein-protein information transfer are important next steps addressable
with computational protein design. Protein-based 3D lattice computers bear great potential for
the future of energy efficient parallel computing.