Soojung Yang

Following the sequence and structure revolutions, predicting functionally relevant protein structure changes at scale remains an outstanding challenge. We introduce BioEmu, a deep learning system that emulates protein equilibrium ensembles by generating thousands of statistically independent structures per hour on a single graphics processing unit (GPU). BioEmu integrates more than 200 milliseconds of molecular dynamics (MD) simulations, static structures, and experimental protein stabilities using new training algorithms. It captures diverse functional motions-including cryptic pocket formation, local unfolding, and domain rearrangements-and predicts relative free energies with 1 kilocalorie per mole accuracy compared with millisecond-scale MD and experimental data. BioEmu provides mechanistic insights by jointly modeling structural ensembles and thermodynamic properties. This approach amortizes the cost of MD and experimental data generation, demonstrating a scalable path toward understanding and designing protein function.

physics-informed equations parameterized with neural networks and provides the total binding affinity of a protein-ligand complex as their sum. We further improved the model generalization by augmenting a broader range of binding poses and ligands to training data. We validated our model, PIGNet, in the comparative assessment of scoring functions (CASF) 2016, demonstrating the outperforming docking and screening powers than previous methods. Our physics-informing strategy also enables the interpretation of predicted affinities by visualizing the contribution of ligand substructures, providing insights for further ligand optimization.

Following the sequence and structure revolutions, predicting the dynamical mechanisms of proteins that implement biological function remains an outstanding scientific challenge. Several experimental techniques and molecular dynamics (MD) simulations can, in principle, determine conformational states, binding configurations and their probabilities, but suffer from low throughput. Here we develop a Biomolecular Emulator (BioEmu), a generative deep learning system that can generate thousands of statistically independent samples from the protein structure ensemble per hour on a single graphics processing unit. By leveraging novel training methods and vast data of protein structures, over 200 milliseconds of MD simulation, and experimental protein stabilities, BioEmu's protein ensembles represent equilibrium in a range of challenging and practically relevant metrics. Qualitatively, BioEmu samples many functionally relevant conformational changes, ranging from formation of cryptic pockets, over unfolding of specific protein regions, to large-scale domain rearrangements. Quantitatively, BioEmu samples protein conformations with relative free energy errors around 1 kcal/mol, as validated against millisecond-timescale MD simulation and experimentally-measured protein stabilities. By simultaneously emulating structural ensembles and thermodynamic properties, BioEmu reveals mechanistic insights, such as the causes for fold destabilization of mutants, and can efficiently provide experimentally-testable hypotheses.

This work considers strategies to develop accurate and reliable graph neural networks (GNNs) for molecular property predictions. Prediction performance of GNNs is highly sensitive to the change in various parameters due to the inherent challenges in molecular machine learning, such as a deficient amount of data samples and bias in data distribution. Comparative studies with well-designed experiments are thus important to clearly understand which GNNs are powerful for molecular supervised learning. Our work presents a number of ablation studies along with a guideline to train and utilize GNNs for both molecular regression and classification tasks. First, we validate that using both atomic and bond meta-information improves the prediction performance in the regression task. Second, we find that the graph isomorphism hypothesis proposed by [Xu, K.; et al How powerful are graph neural networks? 2018, arXiv:1810.00826. arXiv.org e-Print archive. https://arxiv.org/abs/1810.00826] is valid for the regression task. Surprisingly, however, the findings above do not hold for the classification tasks. Beyond the study on model architectures, we test various regularization methods and Bayesian learning algorithms to find the best strategy to achieve a reliable classification system. We demonstrate that regularization methods penalizing predictive entropy might not give well-calibrated probability estimation, even though they work well in other domains, and Bayesian learning methods are capable of developing reliable prediction systems. Furthermore, we argue the importance of Bayesian learning in virtual screening by showing that well-calibrated probability estimation may lead to a higher success rate.

Recently, utilizing reinforcement learning (RL) to generate molecules with desired properties has been highlighted as a promising strategy for drug design. A molecular docking program - a physical simulation that estimates protein-small molecule binding affinity - can be an ideal reward scoring function for RL, as it is a straightforward proxy of the therapeutic potential. Still, two imminent challenges exist for this task. First, the models often fail to generate chemically realistic and pharmacochemically acceptable molecules. Second, the docking score optimization is a difficult exploration problem that involves many local optima and less smooth surfaces with respect to molecular structure. To tackle these challenges, we propose a novel RL framework that generates pharmacochemically acceptable molecules with large docking scores. Our method - Fragment-based generative RL with Explorative Experience replay for Drug design (FREED) - constrains the generated molecules to a realistic and qualified chemical space and effectively explores the space to find drugs by coupling our fragment-based generation method and a novel error-prioritized experience replay (PER). We also show that our model performs well on both de novo and scaffold-based schemes. Our model produces molecules of higher quality compared to existing methods while achieving state-of-the-art performance on two of three targets in terms of the docking scores of the generated molecules. We further show with ablation studies that our method, predictive error-PER (FREED(PE)), significantly improves the model performance.