Kinase-Kernel Models: Accurate Chemogenomic Method for the Entire Human Kinome

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Accurate prediction of drug-target interactions is crucial for drug discovery.
• Kinase inhibitors are a significant class of therapeutics, but screening the entire human kinome is challenging.

Purpose of the Study:

• To develop and validate chemogenomic models for predicting kinase inhibitor activity.
• To interpolate between existing predictive models for comprehensive kinome screening.

Main Methods:

• Utilized chemogenomic kinase-kernel virtual screening models.
• Integrated empirically-trained Profile-QSAR models with IC50 assay data.
• Employed nearest binding-site homologue analysis.

Main Results:

• Successfully interpolated between different modeling approaches.
• Achieved accurate activity predictions for kinases.
• Enabled prediction of activity for the entire human kinome.

Conclusions:

• Chemogenomic models offer a robust method for predicting kinase inhibitor activity.
• The developed models facilitate comprehensive screening of the human kinome.
• This approach advances virtual screening capabilities in drug discovery.