An integrated approach to knowledge-driven structure-based virtual screening.

Area of Science:

• Computational Chemistry
• Drug Discovery
• Structural Biology

Background:

• Structure-based virtual screening (VS) is crucial in drug discovery.
• Existing VS methods often require known ligands to meet specific binding patterns and physicochemical properties.
• Efficiently filtering large compound libraries based on these criteria remains a challenge.

Purpose of the Study:

• To introduce CRAISE, a user-controllable, structure-based VS method.
• To enhance VS by integrating pharmacophore-guided docking with library profiling for property-based filtering.
• To evaluate CRAISE's performance in guiding binding-mode predictions and VS runs.

Main Methods:

• Developed CRAISE, building upon the RApid Index-based Screening Engine (RAISE) approach.
• Implemented pharmacophore-guided protein-ligand docking within CRAISE.
• Utilized library profiles for simultaneous filtering of compounds based on molecular properties.
• Evaluated CRAISE using various pharmacophore hypotheses with different stringency levels.

Main Results:

• CRAISE successfully guides binding-mode predictions, reproducing binding modes below 2 Å for 85% of well-prepared structures when pharmacophore models are appropriate.
• Enrichment of screening results increased, achieving a median AUC of 73%.
• Selectivity was significantly enhanced, leading to up to seven times faster runtimes.

Conclusions:

• CRAISE offers a versatile structure-based VS approach for large-scale assessment of putative ligands.
• The method efficiently integrates pharmacophore matching and property-based filtering.
• CRAISE enhances the accuracy, enrichment, and speed of virtual screening processes.