Related Experiment Videos
Prospective Assessment of Virtual Screening Heuristics Derived Using a Novel Fusion Score.
Dante A Pertusi1, Gregory O'Donnell2,3, Michelle F Homsher2,3
11 Modeling and Informatics, Merck & Co., Inc., West Point, PA, USA.
SLAS Discovery : Advancing Life Sciences R & D
|April 21, 2017
Summary
A novel fusion method for prioritizing drug candidates in high-throughput screening (HTS) significantly improves the identification of active chemical series. This approach enhances iterative screening by combining multiple virtual screening techniques for better drug discovery outcomes.
Area of Science:
- Drug Discovery and Medicinal Chemistry
- Computational Chemistry
- Cheminformatics
Background:
- High-throughput screening (HTS) is crucial for early drug discovery, but screening millions of compounds is resource-intensive.
- Iterative screening processes often employ multiple orthogonal virtual screening methods to prioritize compounds.
- Effective prioritization is essential to maximize the identification of active chemical matter.
Purpose of the Study:
- To introduce and benchmark a novel fusion method for combining prioritizations from orthogonal virtual screening approaches.
- To evaluate the prospective performance of this fusion method across diverse screening campaigns and descriptor spaces.
- To provide guidelines for optimizing iterative screening strategies and compound selection.
Main Methods:
- Developed and applied a novel fusion scoring method to integrate results from multiple virtual screening techniques.
- Benchmarked the fusion approach prospectively on 17 distinct screening campaigns.
- Utilized virtual screening methods across three different descriptor spaces.
- Investigated the impact of weighting similarity and machine-learning scores on enrichment.
- Analyzed the trade-off between initial screening volume and replicate sampling.
Main Results:
- The fusion approach consistently retrieved 15% to 65% more active chemical series compared to any single machine-learning method.
- Optimizing the weighting of similarity and machine-learning scores further increased enrichment by 1% to 19%.
- Prioritizing screening of a larger initial chemical matter set over replicate samples led to the retrieval of more active chemical series.
Conclusions:
- The developed fusion method offers a significant improvement in identifying promising active compounds during iterative drug discovery screens.
- Strategic weighting of different scoring techniques and initial compound selection can substantially enhance the efficiency of virtual screening campaigns.
- These findings provide valuable guidelines for optimizing HTS and virtual screening processes to accelerate the discovery of novel therapeutics.