A local search appproach for transmembrane segment and signal peptide discrimination

Discriminating between secreted and membrane proteins is a challenging task. This is particularly true for discriminating between transmembrane segments and signal peptides because they have common biochemical properties. In this paper, we introduce a new predictive method called LSTranslocon (Local Search Translocon) based on a Local Search methodology. The method takes advantage of the latest knowledge in the field to model the biological behaviors of proteins with the aim of ensuring good prediction. The LS Prediction approach is assessed on a constructed data set from Swiss-Prot database and compared with one of the best methods from the literature.