FlyFinder

FlyFinder is an image-based machine learning tool for analyzing social behavior in Drosophila melanogaster. It uses a trained random forest classifier to accurately segment flies within an observational arena, including clumped or overlapping individuals. After segmentation, FlyFinder calculates inter-fly distances and generates a PDF report with annotated positions, nearest neighbors, and summary statistics. Developed in Python and R, the tool comes with Jupyter Notebooks for streamlined, reproducible analysis.

Jupyter notebooks for Google Colab

• Download all files from the github repo and upload them on google drive
• Configure the data location, output folders and other parameters
  • Users should adjust the following configuration "fly_finder_dir", "train_dir", "test_dir" and "modelName"
• Execute the jupyter notebooks depending on the data (1 panel or 4 panels)on Google colab environment (the Google colab environment makes it easy for users to install the necessary library installation - click "Install dependencies")
• Inspired by https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb

For 4 panels jupyter notebook

• Please use this jupyter notebook for 4 panels

https://github.com/VAI-Metabolism/VAI-Metabolism-FlyFinder-JupyterNotebook-colab/blob/main/FlyFinder_4p_colab.ipynb

• Example - how to use it on Google colab

https://colab.research.google.com/drive/17giUcNKqX7vcTsIr1A2t_yjhTkEqylXI

For 1 panel on github repo

• Please use this jupyter notebook for 1 panel

https://github.com/VAI-Metabolism/VAI-Metabolism-FlyFinder-JupyterNotebook-colab/blob/main/FlyFinder_1p_colab.ipynb

• Example - how to use it on Google colab

https://colab.research.google.com/drive/1VFuBCOjPrKc_VDnfit-Sm3p5gY2iTZ-c#scrollTo=-71mL3byBmQN