Related Experiment Videos
The use of consensus scoring in ligand-based virtual screening
J Christian Baber1, William A Shirley, Yinghong Gao
1Neurocrine Biosciences, San Diego, California 92130, USA.
Journal of Chemical Information and Modeling
|January 24, 2006
Summary
A novel consensus approach combining diverse properties enhances ligand-based virtual screening performance. This method improves data enrichment and consistency compared to single scoring functions in drug discovery.
Area of Science:
- Computational chemistry
- Cheminformatics
- Drug discovery
Background:
- Ligand-based virtual screening (LBVS) is crucial for identifying potential drug candidates.
- Traditional LBVS methods using single scoring functions often face limitations in performance and consistency.
Purpose of the Study:
- To develop and evaluate a new consensus approach for LBVS.
- To improve the performance and reliability of virtual screening by combining multiple scoring methods.
Main Methods:
- Integration of diverse properties: structural, 2D/3D pharmacophore fingerprints, BCUT descriptors, and property-based fingerprints.
- Testing various combination strategies for these properties.
- Utilizing logistic regression and sum ranks for consensus scoring.
Main Results:
- Consensus scoring significantly improves data enrichment compared to single scoring functions.
- Logistic regression and sum ranks demonstrated superior performance in different pharmaceutical applications.
- Consensus methods showed better clustering of active compounds and more consistent ranking across receptor systems.
Conclusions:
- The developed consensus approach offers a robust and consistent strategy for ligand-based virtual screening.
- Combining multiple scoring functions enhances the accuracy and reliability of virtual screening in drug discovery pipelines.