Subchloroplast location prediction via HOmolog knowledge Transfer and feature Selection

Abstract

The accuracy of subchloroplast location prediction algorithms often depends on predictive and succinct features derived from proteins. Thus, to improve the prediction accuracy, this paper proposes a novel SubChloroplast location prediction method, called SCHOTS, which integrates the HOmolog knowledge Transfer and feature Selection methods. SCHOTS contains two stages. First, discriminating features are generated by WS-LCHI, a Weighted Gene Ontology (GO) transfer model based on bit- Score of proteins and Logarithmic transformation of CHIsquare. Second, the more informative GO terms are selected from the features. Extensive studies conducted on three real datasets demonstrate that SCHOTS outperforms three offthe- shelf subchloroplast prediction methods. Copyright © 2013, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.