Prediction of protein-ligand binding affinity via deep learning models

Area of Science:

• Computational chemistry and cheminformatics
• Biophysics and structural biology
• Artificial intelligence in drug design

Background:

• Accurate prediction of protein-ligand binding affinity is vital for drug screening and optimization.
• Traditional computational methods face challenges in predicting binding affinity.
• Recent advances in deep learning (DL) offer new potential for this task.

Approach:

• This work reviews deep learning (DL)-based computational methods for predicting protein-ligand binding affinity.
• It covers traditional methods, DL principles, databases, input representations, and model architectures.
• Challenges and future directions in DL for binding affinity prediction are discussed.

Key Points:

• Deep learning models are increasingly used for predicting protein-ligand binding affinity.
• Limitations include database quality, input representation, and model architecture.
• AlphaFold2's success inspires advancements in DL for drug design.

Conclusions:

• Deep learning holds significant potential for improving protein-ligand binding affinity predictions.
• Addressing current limitations in data and models is key to advancing the field.
• Further research is needed to fully leverage DL for efficient drug discovery.