Hit clustering can improve virtual fragment screening: CDK2 and PARP1 case studies

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Experimental fragment screening is costly and time-consuming.
• In silico methods for fragment screening require further validation.
• Accurate prediction of fragment binding is crucial for drug design.

Purpose of the Study:

• To evaluate existing molecular docking methods for predicting fragment binding.
• To develop a novel virtual fragment screening technique.
• To identify new drug scaffolds, specifically PARP1 inhibitors.

Main Methods:

• Validated Lead Finder docking software using experimental data.
• Analyzed docking of 68 fragments from PDB structures.
• Developed a virtual screening technique involving structural filtration and clustering.

Main Results:

• Achieved RMSD of 1.35 kcal/mol for binding energy prediction.
• Predicted binding poses within 1.5 Å of experimental ones.
• Successfully docked fragments involved in hydrogen bonds and metal coordination in 70-80% of cases.
• Identified two millimolar-scale PARP1 inhibitors with a novel scaffold.

Conclusions:

• Molecular docking can reliably predict fragment binding affinities and poses.
• The developed virtual fragment screening approach is effective for identifying novel inhibitors.
• This method holds potential for drug rediscovery and lead optimization.