Deep behavioral phenotyping of mouse autism models using open-field behavior.

Ugne Klibaite, Mikhail Kislin, Jessica L. Verpeut, Xiaoting Sun, Joshua W. Shaevitz, Samuel S.-H. Wang. Molecular Autism 13, 12 (2022). https://doi.org/10.1186/s13229-022-00492-8 (https://molecularautism.biomedcentral.com/articles/10.1186/s13229-022-00492-8)

This repository contains the analysis code for the above manuscript.

---

MouseMotionMapper

a unique semi-supervised behavioral classification and behavioral labeling pipeline. The framework makes use of deep neural network, PCA, Morlet Wavelet, t-SNE, and k-Mean clustering to define behavior classes. Recent advances in body part tracking Pereira et al. (2019) allow pose estimation of a freely-moving mouse. Here, we adapted behavioral mapping from Berman et al. (2014) for use with the body part position time series.

Sample snippets from each behavioral class generated from movies:

IDLE

GROOMING

SLOW EXPLORE

FAST EXPLORE

REAR

CLIMBING

AMBLE/STEP/TURN

LOCOMOTION

Features and methods included

• Tracking and alignment code for local and cluster
• 2 and 18 body part LEAP models
• Embedding/Clustering code
• Training data + Principal component coefficients
• Code for Fingerprinting
• Code for Embedding new data to behavior space
• Analysis/figure generation code
• Documentation
• Examples of usage

Installation

Git clone the web URL (example below) or download ZIP. It works in Matlab 2019b and above - please submit issues if it's not working. The Image Processing Toolbox, Signal Processing toolbox and the Parallel Processing Toolbox are required.

Change your current working directory to the location where you want the cloned directory to be made.

git clone https://github.com/PrincetonUniversity/MouseMotionMapper

All MATLAB external toolboxes are included in the MouseMotionMapper/utilities subfolder. Just add the subdirectory to the MATLAB Search Path to access all functionality: addpath(genpath('utilities'))

data and meta data

Download the raw data from dataspace.princeton.edu (https://doi.org/10.34770/bzkz-j672)

Usage

The best way to start is by looking at the various demos.

Step 1: Embedding/Clustering

Step 2: Embedding new data to behavior space

Step 3: Fingerprinting

Resources

https://github.com/talmo/leap

https://github.com/gordonberman/MotionMapper

License

This package is licensed under the MIT License. If you use our algorithm and/or model/data, please cite us! Preprint/peer-review will be announced above.

References

1. Gordon J Berman, Daniel M Choi, William Bialek, and Joshua W Shaevitz. Mapping the stereotyped behaviourof freely moving fruit flies.Journal of The Royal Society Interface, 11(99):20140672, 2014.
2. Talmo D Pereira, Diego E Aldarondo, Lindsay Willmore, Mikhail Kislin, Samuel S-H Wang, Mala Murthy, andJoshua W Shaevitz. Fast animal pose estimation using deep neural networks.Nature methods, 16(1):117, 2019.