Albert Keung, Associate Professor, Department of Chemical and Biomolecular Engineering, North Carolina State University
"The Double-Edged Sword of DNA’s Extreme Information Density"
Abstract:
Information technologies are foundational to modern economies, human health, science and engineering, and national security. They have seen major developments in scale and capabilities, the combination of which has driven the rapid acceleration in artificial intelligence. These same developments motivate the need for orders of magnitude improvements in storage and computation, expose challenges in energy and materials usage and resultant climate impacts and resource insecurities, and inspire the potential for completely new transformative capabilities that have yet to be imagined.
Molecular information storage in DNA presents several theoretical advantages if they can be practically realized. DNA can theoretically store orders of magnitude more data per volume and compute for orders of magnitude less energy than current state of the art systems. Here we will present our analysis of how these advantages also present fundamental challenges unique to molecular information systems, and how they must be solved to become practical technologies. We then describe our work addressing these challenges through molecular biology, biochemistry, simulations, and materials science.
In particular, we describe how thermodynamics presents key challenges of undesired non-specific interactions becoming dominant in highly dense systems. We show how we solve this challenge through hierarchical file address systems and through leveraging non-specific interactions to program useful functions like FilePreview/QuickLook. We continue by describing an end-to-end DNA computer that exploits a unique high surface area material for immobilizing DNA while maintaining its accessibility and improving its stability. We demonstrate a 104 TB cm-3 storage density, half-lives over 2 million years at -18C, and we compute simple Sudoku and chess problems in a reuseable manner. We end by describing three unpublished studies. First we describe a system for enzymatic synthesis of modified DNAs. Second, we describe a new fundamental phenomena of DNA molecules in solution that we exploit for controllably obfuscating information. We end by describing a new machine learned model of DNA-DNA interactions that outperforms state of the art models by nearly 30% in accuracy, and we demonstrate the utility of the model by designing a hyperconnected biological network that can efficiently search a database of children’s book characters by their traits.
Bio:
Associate Professor
Goodnight Distinguished Scholar
University Faculty Scholar
Director of Biotechnology, Integrative Sciences Initiative
Associate Director, Biotechnology Program
Department of Chemical and Biomolecular Engineering
Comparative Medicine Institute
Neuroscience and Genetics Programs
Integrative Sciences Initiative
The Keung group engineers cellular and molecular platforms to understand how information is stored and accessed in biological systems. Their work includes engineering human stem cell models of neuroepigenetic disorders, synthetic biology platforms to study and harness the biochemical and dynamic diversity of eukaryotic gene regulation, and application of molecular biology to engineer scalable and extreme density DNA-based information storage and computing systems. His group’s work has been recognized by the AIChE Langer Prize, NIH Avenir Award, ACS Synthetic Biology Young Innovator, CURE Angelman Syndrome Innovation Award, the NSF CAREER Award, and Simons Foundation Pivot Fellowship.