Frequent subsequence-based protein localization

Abstract

Extracellular plant proteins are involved in numerous processes including nutrient acquisition, communication with other soil organisms, protection from pathogens, and resistance to disease and toxic metals. Insofar as these proteins are strategically positioned to play a role in resistance to environmental stress, biologists are interested in proteomic tools in analyzing extracellular proteins. In this paper, we present three methods using frequent subsequences of amino acids: one based on support vector machines (SVM), one based on boosting and FSP, a new frequent subsequence pattern method. We test our methods on a plant dataset and the experimental results show that our methods perform better than the existing approaches based on amino acid composition. © Springer-Verlag Berlin Heidelberg 2006.