DANCE provides an open-source and low-cost approach to quantify aggression and courtship in Drosophila

Abstract

Quantifying animal behavior is pivotal for identifying the neuronal and genetic mechanisms involved. Computational approaches have enabled automated analysis of complex behaviors such as aggression and courtship in Drosophila . However, existing approaches rely on rule-based algorithms and expensive hardware, limiting sensitivity to behavioral variations and accessibility. Here, we present the D rosophila A ggression a n d C ourtship E valuator (DANCE), a low-cost, open-source platform that combines machine learning-based classifiers and inexpensive hardware to quantify aggression and courtship. DANCE consists of six novel behavioral classifiers trained using a supervised machine learning algorithm. DANCE classifiers address key limitations of rule-based algorithms, capturing dynamic behavioral variations more effectively. DANCE hardware is constructed using medicine blister packs and acrylic sheets, with recordings acquired using smartphones, making it affordable and accessible. Benchmarking demonstrated that DANCE hardware performs comparably to high-cost setups. We validated DANCE in diverse contexts, including social isolation vs. enrichment, which modulates aggression and courtship; RNAi-mediated downregulation of the neuropeptide Dsk; and optogenetic silencing of dopaminergic neurons, which promotes aggression. DANCE provides a cost-effective and portable solution for studying behaviors in resource-limited settings or near natural habitats. Its accessibility and robust performance democratize behavioral neuroscience, enabling rapid screening of genes and neuronal circuits underlying complex social behaviors.