JAX Animal Behavior System (JABS): A genetics informed, end-to-end advanced behavioral phenotyping platform for the...

Area of Science:

• Neuroscience
• Behavioral Genetics
• Computer Vision

Background:

• Automated detection of complex animal behavior is crucial for high-throughput studies but faces challenges in adoption by non-computational labs.
• Existing computer vision tools require integrated hardware and software solutions for seamless integration into behavioral neurogenetics research.
• Previous work focused on specific behaviors like grooming and posture in open field arenas.

Purpose of the Study:

• To present the JAX Animal Behavior System (JABS), an integrated platform for automated rodent phenotyping.
• To facilitate data acquisition, machine learning-based behavior annotation, classifier sharing, and genetic analysis.
• To lower the barrier to entry for advanced behavior analysis in neuroscience and genetics.

Main Methods:

• Developed the JAX Animal Behavior System (JABS) with modules for data acquisition (JABS-DA), active learning (JABS-AL), and analysis/integration (JABS-AI).
• Introduced a novel graph-based framework (ethograph) for efficient classifier comparison.
• Integrated downstream genetic analyses (heritability, genetic correlation) with publicly released datasets from 168 mouse strains.

Main Results:

• JABS provides a unified ecosystem for collecting, annotating, and analyzing complex animal behaviors.
• The JABS-AI web application enables classifier sharing and deployment, reducing annotation effort.
• Public release of 168 mouse strain datasets enables genetics-guided classifier selection and analysis.

Conclusions:

• JABS significantly advances automated behavior analysis in neuroscience and behavioral genetics.
• The open-source platform fosters collaboration and reduces technical barriers for researchers.
• Integrated genetic analysis within JABS provides novel insights into behavior-genotype relationships.