Development and Testing of Druglike Screening Libraries.

Area of Science:

• Medicinal Chemistry
• Computational Drug Discovery
• Pharmacology

Background:

• Experimental high-throughput screening (HTS) and in silico virtual screening (VS) are key for identifying drug leads.
• The quality of screening libraries significantly impacts the success of both HTS and VS.
• Existing screening libraries often require improvement for comprehensive drug discovery efforts.

Purpose of the Study:

• To review recent advancements in screening library construction.
• To address the critical need for high-quality screening libraries in drug discovery.
• To compile and evaluate novel druglike screening libraries.

Main Methods:

• Reviewed progress in screening library construction methodologies.
• Compiled screening libraries by performing fingerprint-based similarity searches on 20 million druglike ZINC molecules against known drugs.
• Evaluated library performance using 5847 external active compounds across over 2000 drug targets.

Main Results:

• The compiled screening libraries demonstrated encouraging performance in objective evaluations.
• Achieved 36% retrieval of external actives with a Tanimoto coefficient of 0.75, yielding an enrichment factor of 13.
• Successfully constructed and evaluated drug target family-specific screening libraries.

Conclusions:

• The developed druglike screening libraries are valuable resources for drug discovery.
• These libraries enhance the efficiency and accuracy of virtual screening processes.
• The libraries are publicly available for download, facilitating broader research applications.