Predicting Protein Functions Using Interaction Data
Author Information
Author(s): Cho Young-Rae, Shi Lei, Ramanathan Murali, Zhang Aidong
Primary Institution: State University of New York, Buffalo, NY, USA
Hypothesis
Can integrating protein-protein interaction data with semantic knowledge improve the prediction of protein functions?
Conclusion
Integrating multiple data sources can enhance the prediction accuracy of protein functions.
Supporting Evidence
- The algorithm outperformed competing methods in terms of prediction accuracy.
- Integration of functional knowledge from Gene Ontology improved function prediction.
- The study demonstrated that over 60% of protein interactions do not indicate functional similarity.
Takeaway
Scientists can guess what proteins do by looking at how they interact with other proteins, and using extra information helps make better guesses.
Methodology
The study used a probabilistic framework and leave-one-out cross-validation to predict protein functions based on interaction data and Gene Ontology.
Potential Biases
Potential biases arise from the reliance on existing interaction databases, which may contain erroneous data.
Limitations
The accuracy of predictions is limited by the presence of false positives in interaction data.
Participant Demographics
The study focused on proteins from Saccharomyces cerevisiae.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website