Student Theses and Dissertations

David Jordan

Date of Award

2012

Document Type

Thesis

Degree Name

Doctor of Philosophy (PhD)

RU Laboratory

Leibler Laboratory

Keywords

behavioral variability, Tetrahymena thermophila, motility analysis, digital video microscopy, Jensen-Shannon divergence, artificial selection

Abstract

This work describes our work developing an experimental biological system to study patterns of behavioral variability. We selected motility as the behavior of interest because it is common throughout biology and can be recorded and analyzed relatively easily. We chose to work with a microbe, Tetrahymena thermophila, as a model organism for these studies; it is easy to grow in the laboratory in controlled conditions, has a relatively short generation time, and is large enough to for its motions to be easily imaged. To achieve the imaging, we developed a set of low cost digital video microscopes. Concurrently, we wrote custom software to create trajectories of movements from the recorded movies. Consumer webcams provided high temporal and spatial resolution at low cost, and custom microfluidic devices allowed organisms to be isolated and studied in a well-controlled environment. Simultaneous tracking of multiple individuals, while retaining the identity of each, allowed experiments to span multiple generations. Further, we developed a method of characterizing the swimming behaviors using histograms of linear and angular speeds, which did not rely on explicit modeling or scoring of stereotyped behaviors, and used it to quantitatively measure the similarity between behaviors. These similarities were computed using a relative entropy based metric called the Jensen Shannon divergence. Using this framework, we measured patterns of behavioral changes, both within individual lifetimes and between different individuals in a population. These changes were quantified over time scales that ranged from minutes to hours and even between generations. We measured all of these in a variety of environments, and catalogued the effects of changing the environment. We used the similarity measurements generated from the above analysis to generate a low-(two-) dimensional representation of the behaviors, which led to convenient visualization of the patterns of behavioral change and variability. In addition, we performed experiments using artificial selection that provided evidence that this low dimensional representation may be of biological relevance.

Comments

A thesis presented to the faculty of The Rockefeller University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Permanent URL

http://hdl.handle.net/10209/528

Download

Included in

Life Sciences Commons

Share

COinS