Predicting Protein-Protein Interactions Using Conditional Random Fields
Author Information
Author(s): Hayashida Morihiro, Kamada Mayumi, Song Jiangning, Akutsu Tatsuya
Primary Institution: Bioinformatics Center, Institute for Chemical Research, Kyoto University
Hypothesis
Can conditional random fields combined with mutual information improve the prediction of protein-protein interactions?
Conclusion
The proposed methods using conditional random fields are effective for predicting protein-protein interactions.
Supporting Evidence
- The proposed methods outperformed existing methods like EM and association methods.
- Five-fold cross-validation was used to evaluate the performance of the methods.
Takeaway
Scientists created a new way to predict how proteins interact by looking at their parts and how they work together.
Methodology
The study used computational experiments with protein-protein interaction datasets and conditional random fields to model interactions.
Limitations
The results indicated potential overfitting to training datasets.
Participant Demographics
The study involved protein interaction data from multiple organisms including H. sapiens, D. melanogaster, and C. elegans.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website