Three-dimensional descriptors for aminergic GPCRs: dependence on docking conformation and crystal structure

Area of Science:

• Computational Chemistry
• Medicinal Chemistry
• Structural Biology

Background:

• Three-dimensional (3D) descriptors are crucial for identifying biologically active compounds using ligand- and structure-based methods.
• These descriptors are vital inputs for machine learning (ML) models in quantitative structure-activity relationship (QSAR) studies and activity predictions.
• The accuracy of 3D descriptors and compound orientation representation can significantly impact the predictive power of ML models.

Purpose of the Study:

• To analyze the distribution of 3D descriptors for docked poses of active and inactive compounds across aminergic G protein-coupled receptors (GPCRs).
• To investigate variations in compound conformations influenced by different receptors and crystal structures.
• To understand the factors affecting the consistency of compound orientation within binding sites.

Main Methods:

• Calculation of 3D descriptors for docking poses of active and inactive compounds against aminergic GPCR crystal structures.
• Analysis of descriptor value distributions and their correlation with compound affinity and other molecular properties.
• Focus on variations across different receptors and crystal structures used in docking simulations.

Main Results:

• Compound orientation consistency in binding sites is not strongly correlated with affinity.
• Factors such as the number of rotatable bonds and the specific crystal structure used for docking significantly influence orientation consistency.
• Visualizations of descriptor distributions are available online for further investigation.

Conclusions:

• The choice of crystal structure and ligand flexibility are key factors in achieving consistent compound orientations for docking.
• This analysis provides guidance for selecting appropriate crystal structures in docking studies to enhance ML-based drug discovery.
• Improved selection of docking conditions can lead to better discrimination between active and inactive compounds in predictive modeling.