The MAFDA system can track individual flies within larger groups, identify behaviors, and compare those behaviors to Drosophila genotypes. (Photo credit: Tulane University)
How do you know if a fruit fly is hungry? Ask the computer.
It might sound like a bad dad joke, but it’s a reality at Tulane University. Researchers have developed a new AI tool that can tell whether fruit flies are hungry, sleepy, or singing (yes, fruit flies sing).
Called MAFDA (New Machine Learning-Based Automatic Fly Behavior Detection and Annotation), the system uses cameras and newly developed software to track and identify the complex interactive behaviors of individual flies within larger groups. To do. This allows researchers to compare and contrast the behavior of Drosophila with different genetic backgrounds.
For more than a century, scientists have used the fruit fly’s simple genome and short lifespan to decipher the mysteries of human genetics and immunity. Drosophila melanogaster, fruit flies are known in the scientific world and have won six Nobel Prizes. Drosophila and humans share the same 60% of his DNA.
Previous algorithms had poor accuracy in tracking individual flies within a group, but the MAFDA system facilitated the study of small winged insects.
“Drosophila is something of a pioneer discovering new things, from innate immunity to the chromosome theory of inheritance,” said corresponding authors and professors of biochemistry and molecular biology, Gerald & Flora Joe. – Dr. Wuming Deng, corresponding author of the Mansfield Pilz endowment fund, said. Professor of Cancer Research at Tulane College of Medicine. “Being able to quantify fly behavior is really a step forward in behavioral research.”
Wenkan Liu, a medical graduate student who developed the MAFDA system, said the importance of the platform was “undeniable.”
“It speeds up research, minimizes human error, and provides complex insights into behavioral genetics,” Liu said. “This tool has the potential to be extremely important as it increases reproducibility and opens the door to new explorations in large-scale behavioral analysis.”
MAFDA was developed as part of a recent study that showed that the genes that make flies perceive pheromones—chemicals produced by other fruit flies that are essential for attraction and other processes—are the same genes that control pheromone production in the first place. discovered. . These findings are scientific progresschallenges the current view that separate genes control pheromone production and perception. This finding has broad applications in the fields of human behavioral evolution, metabolism, and sexual dimorphism.
Researchers hope that MAFDA will be used for a variety of purposes in the future. Jie Sun, lead author and postdoctoral fellow at Tulane School of Medicine, said MAFDA could eventually be used to study mice and fish, as well as other insects, and that the system could be used to test drug effects in humans. He said it could be useful for research.
“The more information you give the machine, the more accurately it can discriminate between different behaviors, from courtship to feeding,” says Sun. “This is a very important and meaningful tool.”
MAFDA is already being used in other Tulane research projects, and researchers are working to package the system for use by more scientists at Tulane and around the world.
“That’s the goal,” said Deng Xiaoping. “The original idea was to be able to identify health conditions in flies. It might be too much to ask right now, but we hope this will be used more broadly by the community.” I hope we can move in that direction in the future.”
