Ho, Tchyn Lang Laure;
(2024)
A morphodynamic map of cell division using machine learning.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
UCL_PhD_Thesis_FinalEditedVersion_TLLH.pdf - Other Download (30MB) | Preview |
Abstract
All cells have to undergo dynamic morphological changes throughout the cell cycle, a series of ordered events through which a cell grows and divides into two new cells. Analysis of the observed behaviors is canonically done in subsets of cells by quantifying a set of manually chosen features deemed relevant to the process of interest. Machine learning and image analysis techniques allow to scale this up in an automated and unbiased way, offering descriptions of cell populations as well as individual cells. Nuclear division is the process by which eukaryotes duplicate their nuclear compartment before cell division. In the fission yeast Schizosac- charomyces pombe, this occurs in a closed mitosis form during which an intact nuclear envelope undergoes remarkably reproducible morphologi- cal changes. This work presents a machine learning based framework to explore thousands of fission yeast nuclear division trajectories, with the aim of discovering the rules that govern this essential cell cycle process. First, we performed live-cell fluorescence imaging using markers of the nuclear envelope and microtubules to capture wild-type behavior of nu- clear division as well as aberrant behavior following genetic mutations and treatment with pharmacological components. We then developed an image processing pipeline that produces segmentation masks and single- cell tracks, allowing us to follow individual nuclear divisions in time. Next, we performed unsupervised feature extraction using a deep neural network and showed that the resultant latent features are able to encode two-channel changes that significantly overlap with manually extracted shape features. Finally, when representing nuclear division trajectories in a 2D latent landscape, cells under different conditions are seen travel- ing through different parts of the space, suggesting that this data-driven approach can yield interesting insight into dynamic biological processes. Overall, this proof-of-concept work lays the foundation for the develop- ment of machine learning based prediction in biology.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | A morphodynamic map of cell division using machine learning |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2024. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| Keywords: | self-supervised machine learning, dimensionality reduction, time series analysis, live-cell fluorescence microscopy, fission yeast |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Life Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Life Sciences > Div of Biosciences |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10190780 |
Archive Staff Only
 |
View Item |
