Prediction of kinase-inhibitor binding affinity using energetic parameters

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Molecular modeling

Background:

• Biological activity of molecules is influenced by physicochemical properties and energetic parameters of protein-ligand complexes.
• Predicting molecular activity is crucial for drug discovery and development.

Purpose of the Study:

• To develop a predictive model for biological activity (log IC50) of kinase-inhibitors.
• To identify key molecular descriptors that determine ligand binding affinity.

Main Methods:

• Utilized multiple regression analysis to correlate protein-ligand interaction energy and partition coefficient (logP) with experimental log IC50 values.
• Validated the predictive model on independent datasets of kinase-inhibitor complexes and kinase families.

Main Results:

• Achieved a high correlation coefficient (0.93) in predicting log IC50 values for 25 kinase-inhibitors.
• Demonstrated an average deviation of 0.92 from experimental log IC50 values when tested on 93 kinase-inhibitor complexes.
• Obtained correlation values greater than 0.9 for predicting binding affinities across five kinase families.

Conclusions:

• Protein-ligand interaction energies and partition coefficient values are major determinants of ligand binding affinity.
• The developed regression model shows strong predictive power for kinase-inhibitor activity.
• This approach can guide the design of novel kinase inhibitors with enhanced efficacy.