Molegro Data Modeller to Estimate CDK6 Inhibition

Area of Science:

• Biochemistry
• Computational Chemistry
• Drug Discovery

Background:

• Cyclin-dependent kinase 6 (CDK6) is a key protein target for developing novel anticancer therapeutics.
• Understanding the structural basis of CDK6 inhibition is crucial for drug design.

Purpose of the Study:

• To construct a predictive neural network model for estimating binding affinity to CDK6.
• To explore the potential of machine learning in optimizing drug discovery scoring functions.

Main Methods:

• Utilized Molegro Virtual Docker (MVD) to generate CDK6-ligand complexes and extract relevant features.
• Employed Molegro Data Modeller (MDM) to build regression models using ligand descriptors, energy terms, and scoring functions.
• Developed a neural network model to predict binding affinity, comparing its performance against classical scoring functions like Plant Score.

Main Results:

• The developed neural network model demonstrated superior predictive performance in estimating binding affinity compared to the Plant Score.
• The machine learning approach effectively emulates the exploration of scoring function space for targeted drug design.
• The model provides a robust method for sorting and estimating binding affinity in docking screens.

Conclusions:

• The neural network model offers an effective and targeted approach for CDK6 inhibitor screening.
• This intelligent method enhances the efficiency of drug discovery by improving binding affinity predictions.
• The study provides a valuable tool for researchers in anticancer drug development, with code and data available on GitHub.