DRUMBEAT temporally resolved interpretable machine learning model for characterizing state transitions in protein...

Area of Science:

• Biophysics
• Computational Biology
• Machine Learning

Background:

• Protein conformational transitions are crucial for function but challenging to study mechanistically.
• Existing network models lack the temporal resolution to capture dynamic event sequences.

Purpose of the Study:

• Introduce Dynamically Resolved Universal Model for BayEsiAn network Tracking (DRUMBEAT), a novel machine learning approach.
• Generate interpretable, time-resolved maps of cooperative events in molecular dynamics (MD) trajectories.

Main Methods:

• DRUMBEAT combines a universal graph topology with sliding-window rescoring.
• Applied to Fip35 WW domain folding trajectories to analyze protein dynamics.

Main Results:

• Recovered major folding pathways and identified critical residues.
• Uncovered previously unknown protein features important for conformational transitions.
• Dissected the order and timing of residue contact changes during folding.

Conclusions:

• DRUMBEAT provides a scalable and interpretable framework for studying protein folding dynamics.
• Offers new mechanistic insights into biomolecular conformational transitions.
• Establishes a valuable tool for the mechanistic study of protein dynamics.