Related Experiment Videos
Virtual screening of Chinese herbs with Random Forest
Thomas M Ehrman1, David J Barlow, Peter J Hylands
1Pharmaceutical Sciences Research Division and Centre for Natural Medicines Research, King's College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 9NH, United Kingdom.
Journal of Chemical Information and Modeling
|March 27, 2007
Summary
This study screened Chinese herbal compounds using Random Forest, identifying potential inhibitors for various molecular targets like phosphodiesterases and HIV enzymes. Many herbs showed promise for inhibiting multiple targets, supporting traditional medicine uses.
Area of Science:
- Computational chemistry and pharmacology
- Natural product drug discovery
- Traditional Chinese Medicine (TCM) research
Background:
- Chinese herbal medicine has a long history of use, but scientific validation of its constituents' molecular targets is ongoing.
- Identifying novel inhibitors for therapeutically important targets requires efficient screening methods.
- Understanding the phytochemical basis of TCM efficacy can lead to new drug development.
Purpose of the Study:
- To computationally screen a large database of Chinese herbal constituents for potential inhibitors against key molecular targets.
- To identify compounds that may inhibit targets such as cyclic adenosine 3'-5'-monophosphate phosphodiesterases, protein kinase A, cyclooxygenases, lipoxygenases, aldose reductase, and HIV-related enzymes.
- To explore the potential of Chinese herbs in inhibiting inducible nitric oxide synthase and/or nitric oxide production.
Main Methods:
- Utilized the Random Forest machine learning algorithm for in silico screening of 8264 compounds from 240 Chinese herbs.
- Screened an additional database of 2597 phytochemicals with known target affinities.
- Employed a conservative screening protocol, even with highly unbalanced training data.
Main Results:
- Identified a wide variety of compounds from Chinese herbs with predicted inhibitory activity against multiple molecular targets.
- Found that 83 herb-target predictions were supported by literature evidence.
- Observed that some herbs may contain inhibitors from different phytochemical classes targeting the same molecule, suggesting diverse inhibition strategies.
Conclusions:
- Random Forest is an effective tool for screening natural products, even with imbalanced datasets.
- Chinese herbal medicine represents a rich source of potential inhibitors for various diseases.
- The study highlights the potential for polypharmacology within traditional Chinese herbs, with single herbs inhibiting multiple targets.