The development of a new drug is well known to be very expensive and time-consuming, and accurate drug binding prediction is a prerequisite for fast virtual screening. The latest biological findings observe that the motionless 'lock-and-key' theory is not generally applicable and that changes in atomic sites and binding pose can provide important information for understanding drug binding. However, the computational expenditure limits the growth of protein trajectory-related studies, thus hindering the possibility of supervised learning.

Recently, Dr. Wu from AI Research and Innovation Lab published their latest work to overcome the limitations as mentioned above. They present a novel spatial-temporal pre-training method based on the modified Equivariant Graph Matching Networks (EGMN) named ProtMD (Figure 1), which has two specially designed self-supervised learning tasks: atom-level prompt-based denoising generative task and conformation-level snapshot ordering task to seize the flexibility information inside MD trajectories with very fine temporal resolutions.  ProtMD can grant the encoder network the capacity to capture the time-dependent geometric mobility of conformations along MD trajectories. Two downstream tasks are chosen to verify the effectiveness of  ProtMD through linear detection and task-specific fine-tuning (Figure 2). They observe a huge improvement from current state-of-the-art methods, with a decrease of 4.3% in RMSE for the binding affinity problem and an average increase of 13.8% in AUROC and AUPRC for the ligand efficacy problem. The results demonstrate a strong correlation between the magnitude of conformation's motion in the 3D space and the strength with which the ligand binds with its receptor.

To summarize, their achievements are based on the biological findings that the flexibility of the receptor and ligand is vital in understanding the process of drug binding. Scientists usually rely on the molecular dynamics (MD) simulations to produce their trajectories. Nevertheless, how to exploit this motion information with deep learning remains a key challenge. Their paper proposed a novel spatial-temporal pre-training protocol, dubbed ProtMD. With two delicately designed self-supervised learning tasks, the neural network can capture the time-dependent geometric mobility of conformations along MD trajectories. Their work offers a brand-new perspective of pre-training to learn the flexibility of conformations with MD simulations. It breaks the routine in the molecular representation learning, where molecules are static and immobile, and will shed light on the feature design of deep learning architecture for drug binding. 

This work is supported by Chair Professor Stan Z. Li at the School of Engineering. From the year he joined Westlake University, his team has launched a series of interdisciplinary projects with an aim to accelerate the scientific discovery with AI techniques. ProtMD is a remarkable outcome and if broadly applied, it could offer a powerful pre-training tool that could greatly boost the drug development significantly.