Proteins may exist in or shift among two or more different subcellular locations, and this phenomenon is closely related to biological function. It is challenging to deal with multiple locations during eukaryotic protein subcellular localisation prediction with routine methods; therefore, a reliable and automatic ensemble classifier for protein subcellular localisation is needed. We propose a new ensemble classifier combined with the KNN (K-nearest neighbour) and SVM (support vector machine) algorithms to predict the subcellular localisation of eukaryotic proteins from the GO (gene ontology) annotations. This method was developed by fusing basic individual classifiers through a voting system. The overall prediction accuracies thus obtained via the jackknife test and resubstitution test were 70.5 and 77.6% for eukaryotic proteins respectively, which are significantly higher than other methods presented in the previous studies and reveal that our strategy better predicts eukaryotic protein subcellular localisation.
Key words: Gene ontology, multiple subcellular localisation, K-nearest neighbour, support vector machine, ensemble classifier.
Copyright © 2021 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0