Description:
Market Opportunity
Structure-based virtual ligand screening is emerging as a key paradigm for early drug discovery owing to the availability of high-resolution target structures and ultra-large libraries of virtual compounds. However, to keep pace with the explosive growth of virtual chemical libraries, new approaches to compound screening are needed.
USC Solution
USC researchers developed a new method of structure-based virtual ligand screening in order to identify potential hits for a given target. The system was applied to Enamine REAL virtual library and verified for cannabinoid CB2 receptors screening. The final iteration set was ~460 fold enriched for the 3-component space. Moreover, chemical synthesis and experimental testing of the predicted cannabinoid antagonists yielded a 33% hit rate, with most hits in sub-micromolar range. In a direct comparison these results exceeded those obtained by a standard virtual screening of the Enamine REAL library diversity subset, which required ~100 times more computational resources.
Value Proposition
- AI driven virtual screening platform of virtual libraries commonly used by pharmaceutical companies for hit discovery
- Yields hits with high accuracy that were confirmed in biological screens.
- Requires 100 times less computational resources than competing screening platforms.
Applications
- Screening of virtual libraries
- Identification of virtual hits with high accuracy and speed
- Hits validated in biological screens
Publications
doi: 10.1038/s41586-021-04220-9
Stage of Development:
- Tested on the Enamine virtual library
- Software available
- Available for exclusive and non-exclusive license