Generation of receptor structural ensembles for virtual screening using binding site shape analysis and clustering

Area of Science:

• Computational chemistry and structural biology
• Drug discovery and development

Background:

• Protein flexibility is a critical factor in structure-based virtual screening, yet computationally challenging to model.
• Current methods often simplify protein flexibility, potentially limiting the accuracy of virtual screening.
• Ensemble docking, using multiple receptor conformations, is a viable alternative but requires careful selection of structures.

Purpose of the Study:

• To develop and validate an efficient method for selecting representative receptor conformations for ensemble docking.
• To improve the accuracy and effectiveness of structure-based virtual screening by incorporating protein flexibility.

Main Methods:

• Clustering of receptor conformations based on binding site volume overlaps.
• Ensemble docking calculations using small sets of cluster representative structures.
• Application to crystal structures of cyclin-dependent kinase 2 and HIV protease, and a molecular dynamics ensemble of the androgen receptor.

Main Results:

• A small ensemble of four cluster representative structures effectively accounts for protein flexibility.
• The method consistently yielded high enrichments and diverse active compounds in virtual screening.
• Results from crystal structures and molecular dynamics ensembles were in agreement, demonstrating robustness.

Conclusions:

• Clustering receptor structures by binding site volume is an efficient strategy for ensemble docking.
• This approach enhances the incorporation of protein flexibility into structure-based virtual screening.
• The method represents a significant advancement for improving virtual screening accuracy and drug candidate identification.