The nature of statistical learning theory
The nature of statistical learning theory
On Comparing Classifiers: Pitfalls toAvoid and a Recommended Approach
Data Mining and Knowledge Discovery
Prediction of Protein-Protein Interactions Using Support Vector Machines
BIBE '04 Proceedings of the 4th IEEE Symposium on Bioinformatics and Bioengineering
Kernel methods for predicting protein--protein interactions
Bioinformatics
| Hi-index | 0.00 |
Understanding protein interactions is essential to gain insights into the biological processes at the whole cell level. The high-throughput experimental techniques for determining protein-protein interactions (PPI) are error prone and expensive with low overlap amongst them. Although several computational methods have been proposed for predicting protein interactions there is definite room for improvement. Here we present DomainSVM, a predictive method for PPI that uses computationally inferred domain-domain interaction values in a Support Vector Machine framework to predict protein interactions. DomainSVM method utilizes evidence of multiple interacting domains to predict a protein interaction. It outperforms existing methods of PPI prediction by achieving very high explanation ratios, precision, specificity, sensitivity and F-measure values in a 10 fold cross-validation study conducted on the positive and negative PPIs in yeast. A functional comparison study using GO annotations on the positive and the negative test sets is presented in addition to discussing novel PPI predictions in the pathogen Salmonella Typhimurium.