Selection of GO-Based Semantic Similarity Measures through AMDE for Predicting Protein-Protein Interactions

摘要

Protein-protein interactions (PPI) form the core part of the entire interatomic system for all the living elements. In this article, the role of different Gene Ontology(GO)-based semantic similarity measures in predicting PPIs have been explored. To find out a relevant subset of semantic similarity measures, a feature selection approach is developed with Angle Modulated Differentia] Evolution(AMDE), an improved binary differential evolution technique. In this feature selection approach, SVM classifier is used as a wrapper where different metrics like sensitivity, specificity accuracy and Area Under Curve (AUC) are measured to find the best performing feature subset. Results have been demonstrated for real-life PPI data of yeast.