Developing a high-content imaging dataset to study phenotypic heterogeneity in Saccharomyces cerevisiae.

Candidate: Mahta Jan-Ahmadnejad
Supervisor: Dr. Mojca Mattiazzi Usaj

ABSTRACT

Genetically identical cells grown in the same environment often differ in their cellular morphology, growth characteristics, behavior, and fate. This concept is known as phenotypic heterogeneity, and it can be an advantageous or undesirable feature of biological systems. Several mechanisms can give rise to cell-to-cell variability; however, the role and impact of multiple non-genetic factors on cellular phenotypes remain largely unexplored. The main goal of this project was to prepare a dataset that will enable quantitative exploration of several non-genetic sources of phenotypic heterogeneity in the morphology of five subcellular compartments in the model eukaryote budding yeast, Saccharomyces cerevisiae (S. cerevisiae). Working with wild-type and ~1,200 strains (comprising ~20% of the yeast genome) with mutations in genes involved in major cellular processes, I combined the synthetic genetic array (SGA) method with high- throughput (HTP) microscopy and quantitative image analysis at the single-cell level. In this large-scale project, I specifically focused on analyzing the effects of replicative aging, cell cycle, endoplasmic reticulum stress, and five well-defined environmental conditions. I generated over 2 million raw fluorescence microscopy images, defined 24 mutant and wild-type phenotypes, and prepared training sets for deep learning classifiers that will allow further analysis of the dataset.