Automated Functional Classification of Protein Structures
Author Information
Author(s): Wang Kai, Samudrala Ram
Primary Institution: University of Washington
Hypothesis
Can the FSSA algorithm improve functional classification of proteins compared to traditional methods?
Conclusion
The FSSA algorithm outperforms homology-based approaches for classifying protein functions by utilizing local information.
Supporting Evidence
- About half of the protein structures analyzed belong to multi-functional fold families.
- Approximately one third of functional assignments based on predicted structures are incorrect.
- The FSSA algorithm has been shown to perform better than traditional methods in classifying protein functions.
- The algorithm was evaluated on a larger set of fold families than in previous studies.
Takeaway
This study shows that a new computer program can better guess what proteins do based on their shapes, even when they look similar to other proteins that do different things.
Methodology
The study evaluated the FSSA algorithm on 42 multi-functional fold families using both experimental and predicted protein structures.
Potential Biases
The algorithm may be less sensitive to biases in predicted structures but still relies on the quality of input data.
Limitations
The FSSA algorithm's performance is limited by the reliance on SCOP superfamilies, which may not always accurately reflect protein functions.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website