LarvaTagger: Manual and automatic tagging ofDrosophilalarval behaviour DOI Creative Commons
François Laurent, Alexandre Blanc, L T May

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: March 19, 2024

Motivation As more behavioural assays are carried out in large-scale experiments on Drosophila larvae, the definitions of archetypal actions a larva regularly refined. In addition, video recording and tracking technologies constantly evolve. Consequently, automatic tagging tools for larval behaviour must be retrained to learn new representations from data. However, existing cannot transfer knowledge large amounts previously accumulated We introduce LarvaTagger, piece software that combines pre-trained deep neural network, providing continuous latent representation stereotypical identification, with graphical user interface manually tag train taggers updated ground truth. Results reproduced results an tagger high accuracy, we demonstrated pre-training databases accelerates training tagger, achieving similar prediction accuracy using less Availability All code is free open source. Docker images also available. See git-lab.pasteur.fr/nyx/LarvaTagger.jl .

Language: Английский

Statistical signature of subtle behavioral changes in large-scale assays DOI Creative Commons
Alexandre Blanc, François Laurent, Alex Barbier–Chebbah

et al.

PLoS Computational Biology, Journal Year: 2025, Volume and Issue: 21(4), P. e1012990 - e1012990

Published: April 21, 2025

The central nervous system can generate various behaviors, including motor responses, which we observe through video recordings. Recent advances in gene manipulation, automated behavioral acquisition at scale, and machine learning enable us to causally link behaviors their underlying neural mechanisms. Moreover, some animals, such as the Drosophila melanogaster larva, this mapping is possible unprecedented scale of single neurons, allowing identify microcircuits generating particular behaviors. These high-throughput screening efforts, linking activation or suppression specific neurons patterns millions provide a rich dataset explore diversity responses same stimuli. However, important challenges remain identifying subtle immediate delayed suppression, understanding these on large scale. We here introduce several statistically robust methods for analyzing data response challenges: 1) A generative physical model that regularizes inference larval shapes across entire dataset. 2) An unsupervised kernel-based method statistical testing learned spaces aimed detecting deviations behavior. 3) sequences, providing benchmark higher-order changes. 4) comprehensive analysis technique using suffix trees categorize genetic lines into clusters based common action sequences. showcase methodologies screen focused an air puff, from 280 716 larvae 569 lines.

Language: Английский

Citations

0
Feeding-state dependent modulation of reciprocally interconnected inhibitory neurons biases sensorimotor decisions inDrosophila DOI Open Access

Éloïse de Tredern,

Dylan Manceau,

Alexandre Blanc

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Dec. 26, 2023

Abstract Animals’ feeding state changes behavioral priorities and thus influences even non-feeding related decisions. How is the information transmitted to circuits what are circuit mechanisms involved in biasing decisions remains an open question. By combining calcium imaging, neuronal manipulations, analysis computational modeling, we determined that competition between different aversive responses mechanical cues biased by changes. We found this achieved differential modulation of two types reciprocally connected inhibitory neurons promoting opposing actions. This results a more frequent active type response less frequently protective if larvae fed sugar compared when they balanced diet. The about internal conveyed through homologues vertebrate neuropeptide Y known be regulating behavior.

Language: Английский

Citations

6
Statistical signature of subtle behavioural changes in large-scale behavioural assays DOI Creative Commons
Alexandre Blanc, François Laurent, Alex Barbier–Chebbah

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: May 5, 2024

Abstract The central nervous system can generate various behaviours, including motor responses, which we observe through video recordings. Recent advancements in genetics, automated behavioural acquisition at scale, and machine learning enable us to link behaviours their underlying neural mechanisms causally. Moreover, some animals, such as the Drosophila larva, this mapping is possible unprecedented scales of millions animals single neurons, allowing identify circuits generating particular behaviours. These high-throughput screening efforts are invaluable, linking activation or suppression specific neurons patterns animals. This provides a rich dataset explore how diverse responses be same stimuli. However, challenges remain identifying subtle from these large datasets, immediate delayed suppression, understanding on scale. We introduce several statistically robust methods for analyzing data response challenges: 1) A generative physical model that regularizes inference larval shapes across entire dataset. 2) An unsupervised kernel-based method statistical testing learned spaces aimed detecting deviations behaviour. 3) sequences, providing benchmark complex changes. 4) comprehensive analysis technique using suffix trees categorize genetic lines into clusters based common action sequences. showcase methodologies screen focused an air puff, 280,716 larvae 568 lines. Author Summary There significant gap between architecture selection behaviour generation. have emerged ideal platform simultaneously probing neuronal computation [1]. Modern tools allow efficient silencing individual small groups neurons. Combining techniques with standardized stimuli over thousands individuals makes it relate extracting relationships massive noisy recordings requires development new approaches. suite utilize overarching structure deduce changes raw data. Given our study’s extensive number larvae, addressing preempting potential body shape recognition critical enhancing detection. To end, adopted physics-informed model. Our first group enables within continuous latent space, facilitating detection shifts relative reference second array probes variations sequences by comparing them bespoke Together, strategies enabled construct representations lineage roster ”hit” influence subtly.

Language: Английский

Citations

1
LarvaTagger: Manual and automatic tagging ofDrosophilalarval behaviour DOI Creative Commons
François Laurent, Alexandre Blanc, L T May

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: March 19, 2024

Motivation As more behavioural assays are carried out in large-scale experiments on Drosophila larvae, the definitions of archetypal actions a larva regularly refined. In addition, video recording and tracking technologies constantly evolve. Consequently, automatic tagging tools for larval behaviour must be retrained to learn new representations from data. However, existing cannot transfer knowledge large amounts previously accumulated We introduce LarvaTagger, piece software that combines pre-trained deep neural network, providing continuous latent representation stereotypical identification, with graphical user interface manually tag train taggers updated ground truth. Results reproduced results an tagger high accuracy, we demonstrated pre-training databases accelerates training tagger, achieving similar prediction accuracy using less Availability All code is free open source. Docker images also available. See git-lab.pasteur.fr/nyx/LarvaTagger.jl .

Language: Английский

Citations

0