Related Experiment Videos
Prediction of phosphorylation sites using SVMs
Jong Hun Kim1, Juyoung Lee, Bermseok Oh
1National Genome Research Institute, 5 Nokbun-Dong, Eunpyung-Gu, Seoul, 122-701 Korea. jh7521@ngri.re.kr
Bioinformatics (Oxford, England)
|July 3, 2004
Summary
Predicting protein phosphorylation sites and their kinases using machine learning aids biological research. This method accurately identifies kinase families and groups, aiding functional protein studies and variation analysis.
Area of Science:
- Biochemistry
- Bioinformatics
- Molecular Biology
Background:
- Phosphorylation is a key post-translational modification regulating diverse cellular signal transduction pathways.
- Identifying phosphorylation sites and their associated kinases is crucial for understanding protein function and biological regulation.
Purpose of the Study:
- To develop and evaluate a computational system for predicting phosphorylation sites from primary protein sequences.
- To identify the specific kinase families and groups responsible for phosphorylating these sites.
Main Methods:
- Utilized support vector machines (SVMs) for sequence-based prediction of phosphorylation sites.
- Developed a prediction system, PredPhospho, focusing on four protein kinase families and four kinase groups.
Main Results:
- Achieved high prediction accuracy, ranging from 83% to 95% at the kinase family level.
- Obtained prediction accuracy between 76% and 91% at the kinase group level.
- The PredPhospho system demonstrated effectiveness in predicting phosphorylation events.
Conclusions:
- The developed prediction system, PredPhospho, offers a valuable tool for functional protein studies.
- This system can predict alterations in phosphorylation sites due to amino acid variations across species.
- Computational prediction of phosphorylation enhances understanding of signaling pathways and protein regulation.