Targeting CDK9 with Molegro Virtual Docker.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning in drug discovery

Background:

• Protein-ligand binding affinity prediction is crucial for drug discovery.
• Classical scoring functions in docking programs have limitations in predictive performance.
• Cyclin-dependent kinase 9 (CDK9) is a potential therapeutic target for anticancer drugs.

Purpose of the Study:

• To develop and present a workflow for constructing a neural network model to predict the inhibition of CDK9.
• To integrate computational tools for an enhanced exploration of scoring function space.
• To facilitate the identification of effective regression models for CDK9 inhibition.

Main Methods:

• Utilized Molegro Virtual Docker (MVD) for docking simulations, energy calculations, and descriptor generation.
• Employed Molegro Data Modeller (MDM) to build regression models using MVD-derived features.
• Integrated MVD and MDM with Jupyter Notebooks for a streamlined workflow.
• Sourced binding affinity data from BindingDB and protein structures from the Protein Data Bank.

Main Results:

• Demonstrated a workflow combining MVD, MDM, and Jupyter Notebooks for CDK9 inhibition prediction.
• Successfully constructed a neural network regression model for predicting CDK9 inhibition.
• Facilitated integrated exploration of scoring function space to identify optimal regression models.

Conclusions:

• The presented workflow offers a robust approach for predicting protein-ligand binding affinity, specifically for CDK9.
• This integrated computational strategy enhances the efficiency of identifying potential anticancer drug candidates targeting CDK9.
• The developed models and datasets are publicly available on GitHub for further research and application.