AI and the World's Hard Problems · Life sciences · Nature (computational biology) · 2026-10-26
Design and implementation of an automated Drosophila locomotor assay using computer vision tracking
Drosophila geotaxis assays are valuable for genetics, aging and neuroscience, but scoring is subjective, throughput is low and reproducibility is limited. The authors built an automated platform that analyzes 12 vials in parallel, combining 3D-printed parts, Raspberry Pi video capture and environmental control. A deep learning model segments the vials, and computer vision tracks climbing trajectories and velocity at 60 frames per second. Compared with manual scoring, the pipeline processed data 2.8 times faster with about 800 times higher data density, and it detected reproducible, sex-specific phenotypes in several genotypes.