GLOW: A Workflow Integrating Gaussian-Accelerated Molecular Dynamics and Deep Learning for Free Energy Profiling

Area of Science:

• Computational Chemistry
• Biophysics
• Molecular Dynamics Simulations

Background:

• Understanding biomolecular function requires mapping free energy landscapes.
• Predicting molecular determinants is crucial for drug design and development.
• Advanced simulation techniques are needed to overcome sampling limitations.

Purpose of the Study:

• To introduce a novel workflow, Gaussian-accelerated molecular dynamics (GaMD), deep learning (DL), and free energy profiling workflow (GLOW), for predicting molecular determinants and mapping free energy landscapes of biomolecules.
• To systematically characterize biomolecular mechanisms and identify key reaction coordinates.
• To provide a user-friendly tool for researchers in computational chemistry and biophysics.

Main Methods:

• All-atom GaMD-enhanced sampling simulations were performed on biomolecules.
• Structural contact maps were generated and transformed into images for DL model training using convolutional neural networks.
• DL attention maps identified important structural contacts and system reaction coordinates.
• Free energy profiles were calculated via energetic reweighting of GaMD simulations.

Main Results:

• The GLOW workflow successfully mapped free energy landscapes and identified molecular determinants.
• Application to the adenosine A1 receptor (A1AR) characterized its activation and allosteric modulation.
• GLOW findings aligned with existing experimental and computational data for A1AR.
• New mechanistic insights into A1AR function were obtained.

Conclusions:

• GLOW offers a systematic and powerful approach for mapping biomolecular free energy landscapes.
• The workflow aids in understanding complex receptor dynamics and modulation.
• GLOW provides valuable insights for drug discovery and the design of novel therapeutics.