Tools for Behavioral Phenotyping of C. elegans

Animal behavior is critical to survival and provides a window into how the brain makes decisions and integrates sensory information. A simple model organism that allows researchers to more precisely interrogate the relationships between behavior and the brain is the nematode C. elegans. However, current phenotyping tools have technical limitations that make observing, intervening in, and quantifying behavior in diverse settings difficult. In this thesis, I develop enabling technological systems to resolve these challenges. To address scaling issues in observation and intervention in long-term behavior, I develop a platform for long-term continuous imaging, online behavior quantification, and online behavior-conditional intervention. I show that this tool is easy to build and use and can operate in an automated fashion for days at a time. I then use this platform to understand the consequences of quiescence deprivation to C. elegans health. To quantify complex animal postures, and plant and stem cell aggregate morphology, I develop an app to enable fast, versatile and quantitative annotation and demonstrate that it is both ~ 130-fold faster and in some cases less error-prone than state-of-the-art computational methods. This app is agnostic to image content and allows freehand annotation of curves and other complex and non-uniform shapes while also providing an automated way to distribute annotation tasks. This tool may be used to generate ground truth sets for testing or creating automated algorithms. Finally, I quantify C. elegans behavior using quantitative machine-learning analysis and map the worm’s behavioral repertoire across multiple physical environments that more closely mimic C. elegans’ natural environment. From this analysis, I identified subtle behaviors that are not easily distinguishable by eye and built a tool that allows others to explore our video dataset and behaviors in a facile way. I also use this analysis to examine the richness of C. elegans behavior across selected environments and find that behavior diversity is not uniform across environments. This has important implications for choice of media for behavioral phenotyping, as it suggests that the appropriate media choice may increase our ability to distinguish behavioral phenotypes in C. elegans. Together, these tools enable novel behavior experiments at a larger scale and with more nuanced phenotyping compared to currently available tools.

Level thesis:degree_level

Doctoral

Department dc:contributor.department

Chemical and Biomolecular Engineering

Grantor dc:publisher

Georgia Institute of Technology

Year dc:date.issued

2020

Author dc:creator

• Bates, Kathleen

Advisor dc:contributor.advisor

• Lu, Hang

Committee members dc:contributor.committeemember

• Berman, Gordon
• Goldman, Daniel I.
• McGrath, Patrick
• Styczynski, Mark

• C. elegans
• Behavior phenotyping
• machine learning

Language dc:language.iso

en_US

Handle dc:identifier.uri

http://hdl.handle.net/1853/64968

OAI identifier oai:identifier

oai:repository.gatech.edu:1853/64968