Deep Learning for Behavioural Genomics

Microbiota present a promising and underexploited potential source of novel neuroactive compounds for the treatment of neurological and psychiatric disease in humans. However, due to their diverse and complex interactions with the host nervous system, identifying causal strains and the molecules they produce remains challenging. Here, I address this challenge in the nematode worm, Caenorhabditis elegans, as it is a simple bacterivore with a small nervous system, yet displays a variety of bacteria-influenced behaviours that are observable in the laboratory and governed by conserved neural signalling pathways. I perform phenotypic screening and animal tracking to investigate the behavioural response to various bacteria, in order to highlight behaviour-modifying strains for follow-up analyses to investigate the kinds of neuroactive molecules they produce, and their genetic targets in the worm. Preliminary results find multiple bacterial strains that elicit significant effects on C. elegans behaviour. Follow-up research on these bacteria could lead to the discovery of novel compounds for the molecular control of neural states to treat disease.