Speedy tool to study behaviour in fruit flies

A new machine learning tool, developed in India, could help scientists around the world study behaviour, such as aggression and courtship, in fruit flies without the need for expensive laboratory equipment.

The fruit fly (drosophila melanogaster) is a key organism used in neuroscience, particularly in the study of the genes and brain cells responsible for complex social behaviours.

Traditionally, analysing these behaviours required either expert human observers or costly automated systems.

Now it is hoped that the Drosophila Aggression and Courtship Evaluator (DANCE), an open-source, cost-effective system to study how the brain and the nervous system influence behaviour in humans and animals, could be a quicker and simpler way to do this, also reducing the need for rodent studies in early behavioural research.

Researchers from the Centre for Molecular Neurosciences in Manipal, developed DANCE to identify six key behaviours in male flies, including lunging (aggression) and wing extension, circling, following, attempted copulation and courtship.

The DANCE classifiers closely matched expert human scoring across all behaviours and outperformed previous methods and was able to reproduce results from earlier studies - demonstrating, for instance, how fly behaviour changes with social isolation, food availability, or the genetic manipulation of specific brain cells. 

Pavan Agrawal, senior author of the study available in eLife said: “Its ease of adaptability and portability should also make it useful for studying insect behaviours closer to their natural habitat."