Deep learning-based molecular dynamics simulation for structure-based drug design against SARS-CoV-2.

Area of Science:

• Computational chemistry and structural biology
• Artificial intelligence in drug discovery
• Virology and infectious diseases

Background:

• The COVID-19 pandemic, caused by SARS-CoV-2, necessitates rapid drug development.
• Deep learning (DL) and molecular dynamics (MD) are key computational tools for drug design.
• Existing DL methods face limitations in predicting dynamic binding energy evolution and protein conformational changes.

Purpose of the Study:

• To review recent advancements in DL and DL-based MD simulations for SARS-CoV-2 structure-based drug design (SBDD).
• To highlight how these methods address limitations in protein structure/binding prediction and virtual screening.
• To discuss challenges and future directions for DL-based MD in SBDD.

Main Methods:

• Review of current literature on Deep Learning (DL) applications in drug design.
• Integration of DL with Molecular Dynamics (MD) simulations for complex systems.
• Analysis of DL-based approaches for protein structure prediction, binding affinity, virtual screening, and molecular docking.

Main Results:

• DL and DL-based MD simulations show promise in predicting protein structures and binding interactions for SARS-CoV-2.
• These integrated approaches can dynamically track complex evolution and improve drug virtual screening.
• DL models can overcome traditional neural network deficiencies in predicting interactions with changing protein conformations.

Conclusions:

• DL and DL-based MD simulations offer powerful strategies for accelerating SARS-CoV-2 drug discovery.
• Addressing challenges in dynamic binding energy prediction and conformational changes is crucial for future development.
• Further research into DL-based MD simulations will enhance structure-based drug design efficacy.