Machine learning-based analysis of cell and tissue dynamics driving gastrulation

Supervisors: Dr Rastko SknepnekProf Cornelis J. Weijer

Project Description:

This aim of this highly interdisciplinary PhD project is to combine state-of-the-art imaging, data analysis and sophisticated biophysical modelling to quantitatively analyse the cell behaviours that drive gastrulation in the chick embryo, a model system for human development.  Gastrulation is characterised by large-scale tissue deformations and differential tissue movements that result in the formation of the three primary germ layers, the ectoderm, mesoderm and endoderm in the early embryo. These coordinated tissue deformations derive from the motility and contractility of individual cells interacting with their neighbours, and are additionally highly controlled in space and time through chemical and mechanical cell-cell signalling mechanisms. A key part is to elucidate the cell-cell signalling mechanisms that control these cell behaviours. However understanding the interplay between cell-cell signalling and collective cell behaviours is the major challenge and requires a combination of different approaches such as molecular genetics and cell biology, advanced high-resolution in-vivo imaging, image analysis and extensive mathematical and physical modelling. 
This project will focus on analysis of the changes in cells behaviours of the epithelial epiblast cells driving the formation of the primitive streak, their ingression through the streak and their subsequent movement inside the embryo to form different organs. This will involve live-imaging of distinct cell behaviours of fluorescently labelled cells during normal development and under conditions where candidate cell-cell signalling systems coordinating these behaviours have been disturbed through molecular genetic and direct mechanical manipulation.

To image the over 200,000 cells in the gastrulation stage embryo we have built a highly innovative Fluorescence Light Sheet Microscope that for first time allows detailed live imaging the complex cell behaviours such as shape change, division, movement and ingression in the developing chick embryos where the membranes have been labelled with green fluorescent proteins (Rozbicki et al, 2015). This method is used in combination with high-resolution multi-photon confocal microscopy to study how individual cell behaviours are integrated in the context to produce more complex tissues. These experiments produce enormous quantities of high quality image data (>2TB/experiment) that require a detailed and sophisticated analysis. Therefore this project will make extensive use of advanced computational image processing and data analysis methods including machine and deep learning techniques.
Interpretation of the data will be supported by detailed modelling of gastrulation as collective cell motion using concepts and methods from the physics of soft and active matter (Barton et al., 2017): The PhD candidate will closely collaborate with both the experimental (Weijer) and modelling (Sknepnek) groups on developing, calibrating and refining a novel particle-based model of tissue using the experimental results, and finally applying it to full-scale gastrulation simulations. This highly interdisciplinary project will provide training in advanced cell and developmental biology, molecular genetics, live imaging using advance lightsheet and Multiphoton confocal microscopy as well as provide the opportunity to learn/use advanced computational image processing and mathematical modelling techniques. 


1)    Rozbicki, E. et al. Myosin-II-mediated cell shape changes and cell intercalation contribute to primitive streak formation. Nat Cell Biol 17, 397-408 (2015).
2)    Barton DL, Henkes S, Weijer CJ, & Sknepnek R (2017) Active Vertex Model for cell-resolution description of epithelial tissue mechanics. PLOS Computational Biology 13(6):e1005569.

To apply for this project, please go to this link.