nthndy / cell-comp-analysis

Analysis for cell competition phd project

Analysis for reverse engineering cell competition using multimodal microscopy and deep learning

This repository features various analysis scripts from my PhD project studying cell competition.

Cell Competition

Cell competition is broadly defined as a quality control mechanism that results in less-fit “loser” cells being eliminated from a biological tissue by wild-type "winner" cells. This mechanism impacts a wide variety of different physiological and pathological scenarios, from senescence to tumorigenesis to tissue development.

Methodology

I use time-lapse microscopy to image the behaviours of cellular populations through multiple generations in a competition scenario. Fluorescent markers and quantitative phase interferometry yield a high level of detail into the internal cellular dynamics leading up a certain fate committment, be it mitosis, apoptosis or senescence. Coupling this data with deep-learning powered cellular segmentation and Bayesian multiple-object tracking allows for detailed picture of the chronology of an individual cells cycle: when certain events are initiated and in what circumstances. I aim to use this data to understand what factors play influential roles in triggering cellular decisions. For example, do the “winner” cells that I am studying commit to division in order to crowd out the “loser” cells, or do they simply fill the void left by a “loser” cell that has chosen to apoptose? Is the cellular decision making active or passive?

Scripts

Code here includes:

• Napari scripts for creating neural network training data
• Segmentation analysis scripts
• Plotting scripts for extracting fluorescence and interferometric information
• "Cellular Radial Analysis" scripts for assessing the spatio-temporal distribution of antagonistic cell-cycle events in a cell competition
• Bayesian tracking scripts with personalised parameters for interferometric microscope