Forging a Digital Twin of the Gut-Brain Axis

Our guts shape how we think, and our minds shape how our guts work. Intestinal tracks of all animals, including humans, contain massive numbers of neurons which are essentially brains unto themselves. These enteric nervous systems have direct lines of communication to higher brain centers through specialized nerves (e.g. vagus nerve). Information in these nerves moves both ways and the connections act as information super-highways that sculpt hunger, learning, memory, aggression, locomotion, and even depression and anxiety states. Work in a variety of animals has shown how profoundly this information highway shapes behaviours, but the sheer complexity of the system has been a bottleneck to progress.

One approach to grappling with complexity is to create ‘Digital Twins’. Digital twins are integrative models of physical or biological processes that evolve and grow as new data emerge. In most model organisms, there just isn’t enough information to create detailed twins of the gut-brain axis. However, the necessary data to do this is emerging in fruit fly larvae. The aim of this PhD is to create a digital twin the of the gut-brain axis in fly larvae, which can then evolve digitally and be used to reveal principles of how gut activity shapes higher cognitive functions, and vice-versa.

Pankratz lab at the University of Bonn, working with the Cardona lab at the University of Cambridge is generating a connectome (a complete anatomical map) of the gut-brain axis in fruit fly larvae (1). In parallel, the Pulver lab, working with Prinz lab at Emory University, has developed detailed computer models that simulate the activity of fly neurons (2). Our student will forge a digital twin that incorporates anatomy data from Bonn and neuron simulations from St Andrews/Emory. The student will use anatomical, physiological and computational data to constrain all aspects of the model to ensure that the twin is physiologically realistic. The student will also deploy Artificial Intelligence (AI) and evolutionary algorithms to iteratively tune the model across high dimensional parameter spaces. Next, we will virtually silence and activate cells and circuits within the virtual gut and chart changes in activity in virtual higher brain regions and vice versa. We will use high-dimensional data visualization and dimensionality reduction to screen for ‘power players’ – circuit components with the most power to shape activity across the twin’s nervous system. We will then examine the underlying architecture to determine what functional properties or architectural motifs these power players have. The advantage of this approach is that with a digital twin evolved, refined, and analyzed with cutting-edge computational methods, we will understand not only what anatomical connections are important, but also precisely what patterns of activity across all neurons, at any given time, are most influential.

Digital twinning of the gut-brain axis has the potential to be transformative in both basic research and clinical settings. Here we aim to forge a twin in a less complex animal, which can then grow as new data emerge and serve as a jumping off point for the field.

We acknowledge that research involving international collaboration and intensive computation has environmental costs. We will aim to minimize the environmental impact of our work. Toward this end, our PhD student will create a Sustainability Appendix to their thesis which will aim to provide estimates of the cumulative environmental costs of the PhD in greenhouse gas emissions and outline what was done during the PhD to promote sustainable research practices.

Informal enquiries regarding this scholarship may be addressed to the co-supervisors Stefan Pulver (sp96@st-andrews.ac.uk and Michael Pankratz pankratz@uni-bonn.de).