Expansion of DNA-Encoded Library Hits Using Generative Chemistry and Ultra-Large Compound Catalogs.

Area of Science:

• Drug Discovery
• Computational Chemistry
• Medicinal Chemistry

Background:

• DNA-encoded libraries (DELs) offer powerful hit identification but can be limited by chemical space and hit drug-likeness.
• Generative AI can explore chemical space but struggles with synthesizability and requires initial validated hits.

Purpose of the Study:

• To develop a synergistic methodology combining DEL data and AI for efficient hit expansion.
• To identify novel, drug-like, and commercially available compounds for the chromatin reader protein 53BP1.

Main Methods:

• Leveraging validated DEL data to initialize and bias an AI-powered virtual screening pipeline.
• Expanding initial DEL hits with de novo and purchasable compounds from ultra-large chemical libraries.
• Validating identified compounds using time-resolved fluorescence resonance energy transfer (TR-FRET) assays.

Main Results:

• Identified novel, commercially available hits for 53BP1 from the Enamine REAL Space.
• Three compounds showed TR-FRET IC50 values ≤50 μM; 11 compounds showed IC50 values ≤100 μM.
• AI-nominated hits demonstrated superior chemical diversity, drug-likeness, and purchasability compared to DEL selection.

Conclusions:

• The combined DEL and AI approach provides a streamlined platform for rapid hit expansion.
• This methodology enables efficient discovery of diverse, drug-like, and readily available compounds.
• The synergistic strategy overcomes limitations of DELs and generative AI for accelerated drug discovery.