Improving Protein Localization Prediction with Gene Ontology
Author Information
Author(s): Lei Zhengdeng, Dai Yang
Primary Institution: University of Illinois at Chicago
Hypothesis
Can the performance of protein localization prediction be improved by incorporating Gene Ontology similarity measures?
Conclusion
The study found that using Gene Ontology terms significantly enhances the prediction accuracy of protein subnuclear localizations.
Supporting Evidence
- The new system improved accuracy from 50.0% to 66.5% for single-localization proteins.
- The study utilized a dataset from the Nuclear Protein Database containing over 2000 proteins.
- The best predictive outcome was achieved by summing similarity scores over matched GO term pairs.
Takeaway
This study shows that we can better guess where proteins are located in a cell by looking at their similarities based on Gene Ontology terms.
Methodology
The study used a nearest neighbor classifier combined with Gene Ontology similarity measures to predict protein localization.
Limitations
The system currently predicts only one localization and may overestimate accuracy for proteins with multiple localizations.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website