Machine learning methods for developments of binding kinetic models in predicting protein-ligand dissociation rate...

Area of Science:

• Computational chemistry
• Drug discovery
• Machine learning

Background:

• Protein-ligand binding kinetics are vital for drug potency.
• Current in silico methods struggle with accurate kinetic predictions.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting drug dissociation rate constants.
• To compare the performance of eight different machine learning algorithms.

Main Methods:

• Utilized eight machine learning methods including Bayesian Neural Network (BNN).
• Employed van der Waals/electrostatic interaction energy descriptors.
• Applied models to HSP90 and RIP1 kinase inhibitor case studies.

Main Results:

• The BNN model demonstrated state-of-the-art prediction accuracy for both case studies.
• Achieved high R-squared and low Mean Absolute Error (MAE) values.
• BNN outperformed seven other tested machine learning models.

Conclusions:

• BNN is a highly accurate method for predicting protein-ligand binding dissociation rates.
• This approach enhances in silico prediction of drug kinetic properties.
• Provides a robust tool for drug discovery and development.