Related Experiment Videos
A novel statistical ligand-binding site predictor: application to ATP-binding sites.
Ting Guo1, Yanxin Shi, Zhirong Sun
1Institute of Bioinformatics, MOE Key Laboratory of Bioinformatics, State Key Laboratory of Biomembrane and Membrane Biotechnology, Department of Biological Sciences and Biotechnology, Beijing 100084, China.
Protein Engineering, Design & Selection : PEDS
|April 1, 2005
Summary
A new computational method, the Oriented Shell Model, accurately predicts protein ligand-binding sites. This machine learning approach aids in understanding protein function and identifying potential drug targets.
Area of Science:
- Structural biology
- Computational chemistry
- Bioinformatics
Background:
- Structural genomics rapidly expands the protein 3D structure database.
- Functional characterization often lags behind structural determination.
- Computational methods are needed to predict protein functions, especially ligand-binding capabilities.
Purpose of the Study:
- To develop a novel computational method for predicting protein ligand-binding sites.
- To provide insights into the functional roles of newly determined protein structures.
- To complement existing biochemical research methods.
Main Methods:
- Development of a novel statistical descriptor called the Oriented Shell Model.
- Utilizing distance and angular position distributions of structural and physicochemical features.
- Employing the support vector machine (SVM) as a machine learning classifier.
Main Results:
- The Oriented Shell Model identified 69% of ATP-binding sites in whole-protein scanning.
- High accuracy was observed particularly in eukaryotic proteins.
- The method demonstrates potential for screening ligand-binding-capable protein candidates.
Conclusions:
- The Oriented Shell Model offers a valuable computational tool for predicting protein ligand-binding sites.
- This approach can accelerate functional annotation of protein structures.
- It provides biochemical insights for individual proteins, aiding drug discovery and protein engineering.