Molegro Data Modeller for Machine Learning.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning

Background:

• Protein-ligand interactions are crucial for drug discovery.
• Docking simulations generate vast datasets for analysis.
• Machine learning offers powerful tools for modeling complex biological interactions.

Purpose of the Study:

• To provide a tutorial on building a regression model for predicting cyclin-dependent kinase 2 (CDK2) inhibition.
• To demonstrate the integration of docking simulations with machine learning using Molegro software.
• To explore the utility of scoring function space for developing targeted predictive models.

Main Methods:

• Utilized Molegro Virtual Docker (MVD) for docking simulations of CDK2 inhibitors.
• Employed a support vector machine (SVM) regression model within Molegro Data Modeller (MDM).
• Integrated molecular descriptors, energy terms, and scoring functions from MVD into the machine learning model.

Main Results:

• Successfully built a regression model to predict CDK2 inhibitor binding affinity.
• The model leverages docked poses and specific molecular features for prediction.
• Demonstrated a practical workflow for applying machine learning to docking data.

Conclusions:

• Machine learning, particularly SVMs, can effectively model protein-ligand interactions when integrated with docking simulations.
• The presented tutorial provides a reproducible method for creating targeted predictive models.
• This approach aids in the rational design and discovery of novel kinase inhibitors.