Large Scale Functional Assessment of Cells

Principal Supervisor: Prof. Jens Rittscher (Institute of Biomedical Engineering, Oxford)

Co-Supervisors: Prof. James Johnson, Dr Nicola Beer and Dr Andrew Lowe (Novo Nordisk Research Centre, Oxford)

Background

The release of insulin from pancreatic β cells is necessary for proper glucose homeostasis in mammals. High-throughput time-lapse imaging provides us with the opportunity to study the function of β cells under a broad range of environmental conditions in vitro. For example, with the help of fluorescent molecular markers it is possible to monitor the electrical activity and calcium channel activity of β cells.

The 3D organisation of β cells is also important for the normal regulation of insulin secretion. Recent studies have already revealed functional differences between individual β cells. We aim to develop cellular model systems in 3D that allow us to analyse the topology that regulates population glucose responsiveness. More specifically, we will test if individual cells play distinct roles and if their spatial organisation governs their function.

The analysis of this vast amount of complex time-lapse data is undoubtedly a major bottleneck that needs to be addressed. Relying on human interpretation would restrict the analysis to a small fraction of the available data. Here, latest machine learning methods, including deep learning, will be utilised. Specifically, we aim to address the following research objectives:

Design novel 2D/3D+T feature sets. Novel feature extraction methods are required to summarise the available time-lapse data to identify biologically relevant events and to investigate trends in the data.

Identification of subpopulations. Individual cells will respond differently to external stimuli. In this context is it necessary to support the identification of biologically relevant subpopulations. Machine learning methods will be applied to develop appearance manifolds and learn appropriate distance metrics.

Hierarchical models for cellular behaviour. As the spatial organisation affects the population glucose responsiveness it will be necessary to formulate hierarchical models that can captures this information.

The Novo Nordisk Research Centre Oxford (NNRCO) was launched as a strategic alliance between the University of Oxford and Novo Nordisk, a leading diabetes therapeutics company. The new institute on the Old Road Campus in Headington has world-leading bioinformatics expertise and is in the process of setting up a state of the art high-throughput screening laboratory and. As part of this collaboration NNRCO will provide data sets for the required algorithm development. All experimental studies will be analysed in close collaboration with researchers at NNRCO.