Computationally predicting binding affinity in protein-ligand complexes: free energy-based simulations and machine...

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and development
• Bioinformatics and computational biology

Background:

• Accurate prediction of protein-ligand binding affinities is essential for efficient drug discovery.
• This remains a significant challenge in computational chemistry.
• Various computational methods have been developed to address this problem.

Purpose of the Study:

• To review and compare two major classes of computational methods for predicting protein-ligand binding affinities: free energy-based simulations and machine learning-based scoring functions.
• To discuss recent advancements, particularly in deep learning approaches.
• To comparatively analyze the strengths, weaknesses, and future directions of these methods.

Main Methods:

• Review of thermodynamic cycles for free energy-based simulations.
• Review of feature-representation taxonomies for machine learning-based scoring functions.
• Inclusion of deep learning-based prediction methods that utilize hierarchical feature extraction.

Main Results:

• Free energy simulations and machine learning scoring functions offer potential for accurate binding affinity predictions.
• Deep learning methods leverage hierarchical feature representations for enhanced predictions.
• A comparative discussion of the advantages and disadvantages of each approach is provided.

Conclusions:

• Both free energy simulations and machine learning methods are valuable tools for predicting protein-ligand binding affinities.
• Further research is needed to refine these methods and overcome existing limitations.
• The review highlights key areas for future improvements in computational drug discovery.