Detecting Remote Evolutionary Relationships among Proteins
Author Information
Author(s): Melvin Iain, Weston Jason, Noble William Stafford, Leslie Christina
Primary Institution: NEC Laboratories America
Hypothesis
Can a large-scale algorithm improve the detection of remote evolutionary relationships among proteins?
Conclusion
The ProtEmbed algorithm outperforms traditional methods for detecting remote homology in proteins.
Supporting Evidence
- ProtEmbed achieves superior accuracy to widely used pairwise sequence methods.
- The embedding space can be visualized to provide insights into protein relationships.
- Incorporating structural information improves the performance of the algorithm.
Takeaway
This study created a new way to find similar proteins by looking at their overall structure instead of just comparing them directly.
Methodology
The study used a large-scale algorithm called ProtEmbed to learn an embedding of protein sequences into a semantic space.
Potential Biases
Potential bias from the training data, as it was derived from specific databases.
Limitations
The algorithm was evaluated only on domain sequences and may not generalize to multidomain sequences.
Participant Demographics
The study used protein sequences from the SCOP and ADDA databases.
Statistical Information
P-Value
0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website