Distill: A Suite of Web Servers for Protein Structure Prediction
Author Information
Author(s): Baú Davide, Martin Alberto JM, Mooney Catherine, Vullo Alessandro, Walsh Ian, Pollastri Gianluca
Primary Institution: University College Dublin
Hypothesis
Can we accurately predict protein structural features using a suite of web servers?
Conclusion
The Distill suite provides fast and reliable predictions of protein structural features, allowing for large-scale analysis.
Supporting Evidence
- Porter achieves 79% correct classification on the hard CASP 3-class assignment.
- PaleAle's accuracy exceeds 53% correct 4-class classification.
- BrownAle's classification performance is 16% above a baseline statistical predictor.
- XXStout's performance compares favorably with the best predictors at the latest CASP competition.
Takeaway
This study created a tool that helps scientists figure out how proteins are shaped based on their building blocks, which is really important for understanding how they work.
Methodology
The servers use recursive neural networks trained on large datasets from the Protein Data Bank to predict various structural features of proteins.
Limitations
The predictions may vary in accuracy depending on the availability of homology information.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website