GEESE: Genotype-aware End-to-End Spatio-temporal Embedding for Behavioral Phenotyping

2026-05-22

AI Summary

Traditional behavioral phenotyping relies on handcrafted features, suffering from poor reproducibility and limited scalability. This work proposes an end-to-end deep learning framework that directly learns behavioral representations from 3D pose time-series data, enabling simultaneous behavior classification and genotype prediction without manual feature engineering. The method is the first to uncover genotype-specific behavioral signatures from movement patterns and significantly outperforms conventional approaches across three autism-related genetic models—CNTNAP2, CHD8, and FMR1—demonstrating strong generalization across diverse genetic backgrounds. Built upon a pretrained time-series foundation model, the framework constructs a behavioral manifold and is complemented by HONK, a natural language interface for intuitive interaction with the system.

Abstract

Behavioral phenotyping of genetic animal models currently requires labor-intensive manual feature engineering that limits reproducibility and scalability. We present GEESE, an end-to-end deep learning framework that learns behavioral representations directly from 3D pose dynamics without hand-crafted features. Using a pretrained time series foundation model, we encode movement sequences into a behavioral manifold that supports both behavior classification and genotype prediction. Evaluated across three autism-associated genetic models (CNTNAP2, CHD8, FMR1), our deep learning approach surpasses hand-crafted feature baselines in both tasks, revealing that learned representations capture genotype-specific behavioral signatures. The framework generalizes across genetic backgrounds, and an all-cohort model identifies both genetic background and genotype from movement patterns alone. We further provide HONK, an interactive intelligent tool enabling researchers without programming expertise to perform behavioral phenotyping from pose data through natural language interaction.

Problem

• behavioral phenotyping
• genetic animal models
• manual feature engineering
• reproducibility
• scalability

Innovation

• end-to-end deep learning
• behavioral phenotyping
• genotype-aware embedding
• 3D pose dynamics
• foundation model