Supervised Machine Learning Methods Applied to Predict Ligand- Binding Affinity

Area of Science:

• Computational medicinal chemistry
• Drug discovery and development

Background:

• Predicting ligand-binding affinity remains a challenge in computational medicinal chemistry.
• Machine learning (ML) methods are increasingly used to develop scoring functions for predicting protein-ligand interactions.
• Accurate prediction of binding affinity accelerates early-stage drug development.

Purpose of the Study:

• To review recent advancements in applying ML methods for predicting ligand-binding affinity.
• To highlight the importance of computational approaches in assessing protein-ligand interactions.

Main Methods:

• Focus on computational methods for predicting binding affinity to protein targets.
• Description of major databases for experimental binding constants and protein structures.
• Explanation of methods for evaluating the predictive power of scoring functions.

Main Results:

• Structural information combined with binding affinity data enables the creation of targeted scoring functions.
• Regression analysis facilitates the development of mathematical models for predicting ligand-binding affinities (e.g., inhibition constant, dissociation constant, binding energy).

Conclusions:

• Over 120,000 macromolecular structures are available, alongside evolving binding affinity data, creating a favorable environment for ML-driven scoring function development.
• Scoring functions tailored to specific biological systems demonstrate superior predictive performance compared to general approaches.