Application of support vector machine to three-dimensional shape-based virtual screening using comprehensive
Tomohiro Sato1, Hitomi Yuki, Daisuke Takaya
1RIKEN Systems and Structural Biology Center , 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan.
Journal of Chemical Information and Modeling
|March 20, 2012
Summary
This study combines machine learning with 3D molecular shape analysis to enhance compound screening efficiency. This novel approach improves predictions by using 3D shape similarity profiles, outperforming existing methods when sufficient active compounds are known.
Area of Science:
- Computational Chemistry
- Cheminformatics
- Drug Discovery
Background:
- Traditional 2D similarity methods rely on fingerprints or descriptors, limiting their scope.
- Three-dimensional (3D) molecular shape overlay offers an alternative comparison method but its integration with machine learning is underexplored.
- Improving screening efficiency is crucial for accelerating drug discovery and development.
Purpose of the Study:
- To investigate the application of machine learning, specifically support vector machine (SVM), to 3D molecular shape-based screening.
- To develop and validate new prediction models utilizing 3D similarity profiles for enhanced screening efficiency.
- To assess the performance of 3D shape similarity metrics within a machine learning framework.
Main Methods:
- Combined support vector machine (SVM) with three-dimensional (3D) molecular shape overlay.
- Defined a 3D similarity profile as an array of 3D shape similarities against known active compounds.
- Validated 3D shape similarity metrics (ShapeTanimoto, ScaledColor) from ROCS using known inhibitors of 15 target proteins from the ChEMBL database.
Main Results:
- Machine learning models based on 3D similarity profiles demonstrated stable performance improvements over the original ROCS method.
- Performance enhancement was particularly notable when more than 10 known inhibitors were available as queries.
- The study validated the predictive power of 3D shape similarity metrics within the developed machine learning framework.
Conclusions:
- Combining machine learning with 3D similarity profiles offers a powerful strategy for processing 3D shape information.
- This integrated approach significantly enhances compound screening efficiency compared to conventional methods.
- The findings highlight the potential of 3D shape-based machine learning for accelerating the identification of potential drug candidates.
