Predicting subcellular localizations of membrane proteins in eukaryotes with weighted gene ontology scores

Abstract

Knowing the subcellular locations of proteins is an important step in understanding the protein functions. The experimental methods for determining the protein subcellular localization are usually costly and time consuming. Thus, in the last decade, many computational methods were introduced to determine the protein subcellular location in sillico. Although the physicochemical properties of membrane proteins and globular proteins are different, most of these methods treated the globular proteins and membrane proteins equally. Thus, developing method for determining the subcellular locations for them separately would be necessary. We proposed an algorithm called Weighted Gene Ontology Scores (WGOS) to predict the protein subcellular location of membrane proteins. By comparing the prediction performance of WGOS to MemLoci, which is the only existing method for predicting membrane protein localizations, WGOS performed significantly better than the MemLoci, indicating WGOS might be able to provide better results in practical applications. © 2011 Springer-Verlag Berlin Heidelberg.