Research Projects
DANCE: A Low-Cost, Machine Learning-Based Platform for Quantifying Drosophila Social Behaviors
Understanding animal behavior is key to uncovering the neural and genetic mechanisms underlying complex social interactions. However, existing tools for behavioral quantification are often expensive, rigid, and inaccessible, limiting their applicability in diverse research settings. To address these challenges, we developed DANCE (Drosophila Aggression and Courtship Evaluator), an open-source, low-cost platform that integrates machine learning-based behavioral classifiers with affordable hardware to quantify aggression and courtship behaviors in Drosophila.
DANCE incorporates six supervised machine learning classifiers, trained using JAABA, to robustly identify key aggressive and courtship behaviors. These classifiers surpass traditional rule-based algorithms by capturing subtle variations in behaviors.
The DANCE hardware is ultra-low-cost (<30 Rupees), constructed from repurposed materials such as medicine blister packs and acrylic sheets, enables high-quality behavioral recordings using standard Android smartphones. Benchmarking demonstrates that DANCE performs comparably to high-end systems costing over $3500.
We validated DANCE in multiple experimental contexts, including social isolation versus enrichment, RNAi-mediated gene knockdown, and optogenetic manipulation of dopaminergic neurons. These studies confirm its ability to detect subtle behavioral changes with high sensitivity, underscoring its potential for rapid genetic and neural circuit screening.
DANCE is fully open-source, ensuring accessibility for researchers worldwide. Our workflow includes comprehensive setup guides, training videos, and pre-trained behavioral classifiers to facilitate easy adoption.
Publication: Yadav R S P#, Dey P#, Ansari F, Kottat T, Prabhu PP, Vasam M, Ayyangar S, Bhaskar BS, Prabhu K, Ghosh M, Agrawal P*, DANCE: An open-source analysis pipeline and low-cost hardware to quantify aggression and courtship in Drosophila. eLife (in review, 2025) (link). (#Co-first authors).
Explore DANCE on our GitHub page: https://github.com/agrawallab/DANCE
Figure: Aggression and courtship behaviors of Drosophila fruit flies analyzed with the DANCE method, demonstrating differences from existing methods (Adapted from Yadav & Dey et al., eLife, 2025).
Social isolation-induced aggression mediated by neuropeptide Drosulfakinin
Social isolation is a form of 'passive stressor' which causes long term deleterious effects in humans, e.g., prisoners kept in solitary confinements show increased violence and sleep disturbances. Interestingly, social isolation in Drosophila has similar effects on behaviors to those seen in humans and rodents including increased aggression and decreased sleep duration. In this project, we investigated how social isolation influences gene expression in the brain leading to subsequent increase in aggression. Using RNA-seq from heads of flies undergone social isolation vs. social enrichment, we identified 90 differentially expressed genes. One of these genes is Drosulfakinin (Dsk) that encodes for a neuropeptide, which is homologous to the vertebrate Cholecystokinin (CCK). CCK regulates many aspects of aggression, anxiety and social behavior in mammals. Intravenous injection of CCK in humans reliably induces panic attacks and is often used to screen anxiolytic drug candidates. We investigated Drosophila aggressive behavior with high-throughput machine vision based assays and found that Dsk knockdown in small number of Dsk secreting neurons or Drosophila insulin-like-peptide-producing cells (Dilp2), increased isolation-induced male-male aggression (Figure 1). By targeted genetic knockdowns of Dsk ligand and its receptors; increasing and decreasing neural firing of Dsk producing cells, we found that Dsk signaling has a U-shaped dependence on social isolation induced aggressive behavior in Drosophila.
Figure 1 : Dsk knockdown increases social isolation-induced aggression independently of overall activity levels. Dsk levels were reduced by driving the expression of UAS-Dsk-RNAi with Dsk-GAL4 or Dilp2-GAL4. These drivers overlap in the pars intercerebralis region (white arrowhead, A and B). RNAi-mediated Dsk knockdown increased aggression in SH flies relative to the controls without RNAi insert (C, and E). Dsk knockdown significantly reduced the overall daytime activity of SH males (D, and F) suggesting that increase in aggression upon Dsk knockdown is not merely due to overall increase in activity. Adapted from: Agrawal et al., 2019, https://doi.org/10.1101/646232
Publication: The neuropeptide Drosulfakinin regulates social isolation-induced aggression in Drosophila; Agrawal P*, Kao D, Chung P, Looger LL; Journal of Experimental Biology, 2020, 223: jeb207407. Also in- BioRxiv, 2019. (*Corresponding author)
Cell type specific behavioral epigenetics in Drosophila
Epigenetic mechanisms, including histone post-translational modifications such as acetylation, methylation, phosphorylation etc. play a fundamental role in regulating gene expression in the brain and regulating behaviors. Brain is the most heterogeneous tissue in the body and therefore, whole genome epigenetic and transcriptional data obtained from subsets of brain tissue can be quite noisy. This is even more challenging for small model organisms such as Drosophila, where manually dissecting subsets of brain regions for epigenomic analysis is not possible. Because of this issue cell-type specific behavioral epigenetic studies were not possible in Drosophila.
To address this issue, we developed a modification of the INTACT method (Henry et al., 2012) in Drosophila, called 'mini-INTACT', in which neurons of interest are labeled by expressing a tagged-GFP, using UAS-GAL4 system. The tag localizes GFP to the inner nuclear membrane which can be exposed during homogenization and nuclei of interest can be affinity purified for epigenetic and transcriptional profiling (Figure 2).
Figure 2: mini-INTACT method affinity purifies cell-type-specific nuclei for epigenetic analysis in Drosophila. Adapted from: Agrawal et al., BMC Biol, 17, 30, 2019.
Modifications made in mini-INTACT protocol reduced required input material by ~100-fold, enabling use of Drosophila as a model for cell-type specific behavioral epigenetics. Therefore, mini-INTACT can be used to study epigenetic changes across various paradigms including stress, drugs of abuse and neuro-degenerative disorders.
Purity of mini-INTACT (~97%) is comparable or better than FACS based purification and has additional advantages: (1) Unlike FACS, potential cellular stress is circumvented since animals expressing tagged-GFP can be subjected to behavioral modification and flash frozen for subsequent nuclei purification. (2) Animals are collected in a restricted temporal window avoiding potential changes in epigenetic and transcriptional state.
We used mini-INTACT to purify nuclei from dopaminergic neurons, a rare cell type, which comprises less than 0.1% of fly brain neurons. Using ChIP-seq on mini-INTACT purified dopaminergic nuclei, we identified epigenetic changes in Drosophila males that were socially isolated and socially enriched for four days. Social experience altered the epigenetic landscape in clusters of genes involved in transcription and neural function (Figure 3, A). Some of these alterations could be predicted by expression changes of four transcription factors and the prevalence of their binding sites in several clusters. These transcription factors were previously identified as activity-regulated genes (ARGs), and their knockdown in dopaminergic neurons reduced the extent of sleep modulated by social experience, denoted as delta Sleep (Figure 3, B).
Figure 3: (A) Epigenetic landscape of genes expressed in dopaminergic neurons is modulated by social experience. Heat map of eight groups identified by k-means clustering of the change in whole gene average mark levels and mRNA levels between GH and SH males. Red lines show genes whose marks or mRNA was higher in GH than SH males, blue lines show those that were higher in SH than GH males. (B) Knockdown of activity regulated genes (ARGs) by driving RNAi in dopaminergic neurons (TH-GAL4) affected social effects on daytime sleep. Controls carried empty vectors without RNAi hairpin and TH-GAL4. Adapted from: Agrawal et al., BMC Biol, 17, 30, 2019.
Publications: 1. Enabling cell-type-specific behavioral epigenetics in Drosophila: a modified high-yield INTACT method reveals the impact of social environment on the epigenetic landscape in dopaminergic neurons; Agrawal P*, Chung P, Heberlein U, Kent F C*; BMC Biology; 2019, 17, 30. Also in- BioRxiv, 2018. (*Co-corresponding author).
2. Kent, C*., Agrawal, P*. Regulation of Social Stress and Neural Degeneration by Activity-Regulated Genes and Epigenetic Mechanisms in Dopaminergic Neurons. Molecular Neurobiology, 57, 4500–4510 (2020). https://doi.org/10.1007/s12035-020-02037-7 (*Co-corresponding author).