Predicting Protein Binding Sites Using Interface Propensities
Author Information
Author(s): Dong Qiwen, Wang Xiaolong, Lin Lei, Guan Yi
Primary Institution: School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China
Hypothesis
Can residue-level and profile-level interface propensities improve the prediction of protein binding sites?
Conclusion
The binary profile interface propensities significantly enhance the accuracy of protein binding site predictions compared to residue interface propensities.
Supporting Evidence
- The binary profile interface propensities improved binding site prediction performance by about ten percent in terms of precision and recall.
- Residue interface propensities showed minor differences among the four types of protein complexes.
- Machine learning methods, particularly support vector machines, were effective in classifying interface residues.
Takeaway
This study shows that using special patterns from protein sequences can help scientists find where proteins bind to other molecules better than before.
Methodology
The study used support vector machines to analyze residue interface propensities and binary profile interface propensities for predicting protein binding sites.
Limitations
The improvement in prediction accuracy with residue interface propensities was negligible compared to binary profile propensities.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website