GPCRTree: A New Tool for Classifying GPCRs
Author Information
Author(s): Matthew N Davies, Andrew Secker, Mark Halling-Brown, David S Moss, Alex A Freitas, Jon Timmis, Edward Clark, Darren R Flower
Primary Institution: The Jenner Institute, University of Oxford
Hypothesis
Can we develop a more accurate method for classifying G protein-coupled receptors (GPCRs) based on their sequences?
Conclusion
GPCRTree is significantly more accurate than existing GPCR prediction servers at classifying GPCRs.
Supporting Evidence
- GPCRTree produced accuracies of 97% at the Class level, 84% at the Sub-family, and 75% at the Sub-Subfamily level.
- It is the first server to implement an alignment-independent representation of protein sequences.
- GPCRTree is currently the most accurate publicly-available server for GPCR sequence classification.
Takeaway
GPCRTree is a computer tool that helps scientists figure out what different proteins do by looking at their sequences, and it's better than other tools at this job.
Methodology
An alignment-free classification approach using a dataset of GPCR sequences and a selective top-down classifier.
Potential Biases
Different classifiers may have biases that affect classification accuracy.
Limitations
The classification model does not include Class F GPCRs due to insufficient data.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website