SeqFEATURE: A Tool for Protein Function Annotation
Author Information
Author(s): Wu Shirley, Liang Mike P, Altman Russ B
Primary Institution: Stanford University
Hypothesis
Can SeqFEATURE improve protein function annotation using structural representations of sequence motifs?
Conclusion
SeqFEATURE effectively models protein functions and shows improved performance over existing methods, especially when sequence and structural similarities are low.
Supporting Evidence
- SeqFEATURE models show 77% have an AUC greater than 0.8.
- 82% of models have sensitivity greater than 0.5 at the default cutoff.
- SeqFEATURE predicts 60% fewer false positives compared to PROSITE.
Takeaway
SeqFEATURE helps scientists understand what proteins do by looking at their shapes instead of just their sequences, making it easier to find functions for new proteins.
Methodology
SeqFEATURE builds a library of 3D functional site models from PROSITE motifs and evaluates their performance through cross-validation.
Potential Biases
Potential bias due to the selection of training sets and the reliance on curated databases.
Limitations
The method may be limited by the quality of the training sets and the reliance on existing databases for functional annotations.
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