Related Experiment Videos
Recursive median partitioning for virtual screening of large databases.
Jeffrey W Godden1, John R Furr, Jürgen Bajorath
1Department of Computer-Aided Drug Discovery, Albany Molecular Research, Inc. (AMRI), 21 Corporate Circle, Albany, New York 12212-5098, USA.
Journal of Chemical Information and Computer Sciences
|January 28, 2003
Summary
The recursive median partitioning (RMP) method enhances virtual screening by efficiently identifying active compounds. This approach significantly improves hit rates and reduces candidate molecules in large databases.
Area of Science:
- Computational chemistry
- Cheminformatics
- Drug discovery
Background:
- The median partitioning (MP) method was previously developed for diversity selection and compound classification.
- MP utilizes property descriptors and database medians to classify molecules.
Purpose of the Study:
- To extend the MP methodology for application in virtual screening.
- To develop a recursive MP (RMP) approach for efficient identification of active compounds.
Main Methods:
- The RMP approach recursively partitions large compound databases based on descriptor combinations.
- A genetic algorithm (GA) facilitates descriptor selection for optimal copartitioning of active molecules.
- The RMP method was applied to five diverse biological activity classes using a database of ~1.34 million molecules.
Main Results:
- RMP analysis achieved hit rates up to 21%, varying by biological activity class.
- The method led to an average ~3600-fold improvement over random selection.
- RMP effectively reduced candidate molecules in partitions enriched with active compounds.
Conclusions:
- The recursive median partitioning (RMP) method is effective for virtual screening.
- RMP significantly enhances the efficiency of identifying biologically active compounds from large databases.
- This methodology offers a substantial improvement over traditional random selection methods.