PETIMOT introduces a revolutionary framework for predicting protein motion patterns from limited experimental data, addressing a critical gap in structural biology.

• Leverages SE(3)-Equivariant Graph Neural Networks to generate continuous representations of protein movements
• Develops a task-specific loss function that preserves data symmetry across scaling and measurements
• Creates accurate motion models from sparse experimental observations
• Enables better understanding of protein function through improved conformational analysis

This research significantly advances our ability to model protein dynamics, essential for drug discovery, understanding disease mechanisms, and developing targeted therapeutics.

PETIMOT: A Novel Framework for Inferring Protein Motions from Sparse Data Using SE(3)-Equivariant Graph Neural Networks