Predicting Protein Interactions from Sequence Information
Author Information
Author(s): Betel Doron, Breitkreuz Kevin E, Isserlin Ruth, Dewar-Darch Danielle, Tyers Mike
Primary Institution: Samuel Lunenfeld Research Institute, Mt. Sinai Hospital, Toronto, Ontario, Canada
Hypothesis
Can we predict novel protein interactions using only sequence information?
Conclusion
The study demonstrates that new protein interactions can be predicted exclusively from sequence information.
Supporting Evidence
- The method predicted 18,459 interactions between 2,313 proteins.
- 609 predicted interactions were confirmed with experimental evidence.
- The predictions were based solely on sequence information.
Takeaway
Scientists found a way to guess how proteins stick together just by looking at their sequences, like figuring out how puzzle pieces fit without seeing the whole picture.
Methodology
The study used a method called D-MIST to predict protein interactions based on sequence profiles derived from structural data.
Potential Biases
The analysis may be biased due to underrepresentation of certain protein structures in existing databases.
Limitations
The method is limited by the availability of detailed binding information and may not account for all possible domain-binding sequences.
Participant Demographics
The study focused on yeast proteins.
Statistical Information
P-Value
1.0 × 10−13
Statistical Significance
p = 1.0 × 10−13
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website