As neuroscience seeks to understand larger, more complicated systems and behaviors, we will require neural recording techniques that can monitor the activity of neurons across the whole brain. The unparalleled data-storage capabilities of DNA, combined with fast, genetically encodable DNA polymerase-based sensors, would allow for "ticker tape"-like recordings of neurons across the whole brain, allowing for new experimental paradigms in neuroscience. Such a system does not yet exist, in neuroscience or otherwise. The work contained here represents early computational and biological demonstrations of such a DNA polymerase-based molecular recorder. We develop an algorithm that solves a central statistical problem for this and other types of molecular recorders, namely the lack of true temporal indexing. This algorithm improves upon the parallelizability of dynamic time warping-class algorithms, and allows for likelihood-based estimation of neural properties in a simulated motor control experiment. We also use this algorithm to determine parameters for DNA polymerase-based molecular recorders and experiments that allow for decoding. We demonstrate DNA polymerase-based recording in a simple in vitro setting, recording several bits of information onto single strands of DNA. Through this work, we describe several benchmarking techniques that can be used to characterize critical characteristics of molecular recording systems more generally. Through these computational and biological aims, I provide a broad development toolkit, as well as an early proof-of-concept, for DNA polymerase-based molecular recorders.

Last modified

• 01/10/2019

Creator

• Thaddeus Rudas Cybulski

DOI

• https://doi.org/10.21985/N2FN1G

Subject

• Interdepartmental Neuroscience Program (NUIN)

Keyword

• Next-generation sequencing
• DNA Polymerase
• Dynamic Time Warping
• Molecular recording
• Brain activity mapping

Date created

• 2017-01-01

Resource type

• Dissertation

Rights statement

• In Copyright