FlyBowl Behavior Data and JAABA Classifiers for Drosophila

Aggregating raw FlyBowl experiment data and 17 trained JAABA behavior classifiers for Drosophila. It was prepared for the Neurons Can Fly outreach program at the University of Michigan and includes data for GAL4 activation lines with large effect sizes for chase and wing extension behaviors.

Use Cases

• Apply the 17 trained JAABA classifiers to detect and quantify Drosophila behaviors like chase and wing extension in new FlyBowl videos.
• Analyze the effect sizes of specific GAL4 activation lines (e.g., R82E08, R15A01) on chase behavior compared to pBDPGAL4U control experiments.
• Use the provided experiment directories to retrain or validate JAABA classifiers for specific Drosophila behaviors.
• Compare behavioral phenotypes between different genetic lines (e.g., R71G01, R32C09) for wing extension behavior.

Strengths

• Includes 17 trained JAABA behavior classifiers for automated analysis.
• Contains data for specific GAL4 lines with documented large effect sizes on behavior.
• Designed for educational use in a structured outreach program.

Limitations

• The dataset scope is focused on specific genetic lines and behaviors, limiting generalizability.
• Data volume, file formats, and specific row/column counts are not documented.

Provenance

Source

Harvard Dataverse, contributed by author Kristin Branson.

Collection Method

Raw data from Fly Olympiad FlyBowl experiments and trained JAABA classifiers.

Freshness

Last updated on 2026-02-13.