Abstract

Synthetic Biology provides an avenue for reengineering the molecular machinery that make up cells. It has the potential of becoming a significant driver for discovery of new therapies and diagnostic methods. In fact, advances in molecular biology have made it easier to create large pools of edited cells, but there is a technological bottleneck to screen these cells to capture their phenotypes and link them to genotypes. Conventional screening technologies like well based structured arrays and Fluorescence-activated cell sorting (FACS) provide a means to screen genetically edited cells, but their current limitations prevent capturing dynamic phenotypes from mixed populations of edited cells. Microfluidic technologies provide alternatives that can be combined with timelapse microscopy to capture phenotypes. Paired with other techniques, these devices can provide ways to genotype mixed populations in situ and externally with single cell resolution. This work involves one of such techniques referred to as Single Cell Isolation Following Timelapse (SIFT), used to screen mixed libraries of synthetic oscillators. However, it is currently limited to Escherichia coli (E.coli) and further research is needed to adapt it to mammalian cells. As such, this thesis presents our implementation of the technique for screening reengineered E. coli cells in combination with an existing machine learning segmentation method referred to as Deep Learning Time-lapse Analysis (DeLTA). Similarly, this work features preliminary results to extend SIFT to Jurkat cells (human leukemic T cell line ). In brief, the presented work involves the implementation of a microfluidic set-up to screen mixed populations of edited cells.