Method for Prediction of Protein-Protein Interactions in Yeast using Genomics/Proteomics Information and Feature Selection pp. 77-102
Authors: (J.M. Urquiza, I. Rojas, H. Pomares, L. J. Herrera, Dept. of Computer Architecture and Computer Technology, University of Granada, Granada, Spain)
Abstract: Nowadays,one of the most important goals of Proteomics is the prediction of protein-protein interactions (PPIs), whose knowledge is vital for all biological processes. In the present paper we propose an approach to the prediction of protein-protein interactions in yeast based on the well-known paradigm of Support Vector Machines(SVM)for classification and features election methods using Genomics/Proteomics information from the main databases. In order to obtain higher values of specificity and sensitivity for our predictions, we took a high reliable set of positive and negative examples for the construction of the SVMmodel. We then extracted a set of proteomic/genomic features from these examples and also introduced a similarity measure in the calculation of the features,that allows us to improve the prediction capability of our model. In the analysis of the results, we also applied our approach to invitro data sets, obtaining high accuracy classifications. Our final SVM classifiers obtain a low error rate in the prediction for each pair of proteins of several data sets for both invitro and in silico methodologies.