A study of network-based kernel methods on protein-protein interaction for protein functions prediction

Abstract

Predicting protein functions is an important issue in the post-genomic era. In this paper, we studied several network-based kernels including Local Linear Embedding (LLE) kernel method, Diffusion kernel and Laplacian Kernel to uncover the relationship between proteins functions and Protein-Protein Interactions (PPI). We first construct kernels based on PPI networks, we then apply Support Vector Machine (SVM) techniques to classify proteins into different functional groups. 5-fold cross validation is then applied to the selected 359 GO terms to compare the performance of different kernels and guilt-by-association methods including neighbor counting methods and Chisquare methods. Finally we made predictions of functions of some unknown genes and verified the preciseness of our prediction in part by the information of other data source.