Introduction: Molecular dynamics (MD) is a powerful computational technique for simulating biomolecular dynamics at an atomistic level. Gaussian accelerated molecular dynamics (GaMD) is an unconstrained enhanced sampling technique that allows for accurate free energy calculations. Machine learning (ML) has been applied to enhance MD simulations and facilitate the simulation analysis of biomolecules, particularly G-protein-coupled receptors (GPCRs). However, ML has not been combined with MD for modeling of GPCR allosteric modulation. In this work, we introduce a GaMD, Deep Learning (DL) and free energy profiling workflow (GLOW) to predict molecular determinants and map free energy landscapes. The GLOW workflow can be downloaded at http://miaolab.org/GLOW.

Methods: In GLOW, GaMD and DL are integrated to identify important reaction coordinates (RCs) and map free energy profiles of biomolecules. First, GaMD simulations are performed on target biomolecules. Since simulation trajectories are collections of static PDB snapshots, the residue contact map of each simulation frame can be calculated and transformed into grayscale images. 80% of the resulting images are randomly selected for training, while the rest are used for validation. The specialized type of neural network for image classification, 2D convolutional neural network (CNN), is employed to classify the residue contact maps of target biomolecules. The model architecture consists of four convolutional layers, followed by three fully connected (dense) layers. “ReLU” activation was used for all layers in the 2D CNN, except the classification layer, where “softmax” activation is used. A maximum pooling layer is added after each convolutional layer. Finally, important residue contacts are identified by classic gradient-based pixel attribution. The free energy profiles of the RCs are calculated through reweighting of GaMD simulations to characterize the biomolecular systems.

Results: We have successfully demonstrated GLOW on characterization of activation and allosteric modulation a GPCR, using the adenosine A₁ receptor (A₁AR) as a model system. GLOW findings are highly consistent with previous experimental and computational studies of the A₁AR, while also provide further mechanistic insights into the receptor function. GLOW has, for the first time, enabled combining ML and MD for modeling of GPCR allosteric modulation. It provides a systematic approach to map free energy landscapes of biomolecules.

Apurba Bhattarai

PhD Candidate, University of Kansas

Appearances