Prediction of signal peptides using scaled window

Cells use a ZIP code system to sort newly synthesized proteins and deliver them wherever they are needed: into different internal compartments called organelles or even out of the cell altogether. One of the most essential features of the ZIP code system is the signal sequence or "address tag," which is originally present in the N-terminal part of the protein and is trimmed away by the time it is secreted. Owing to the importance of signal peptides for understanding the molecular mechanisms of genetic diseases, reprogramming cells for gene therapy, and constructing new drugs for correcting a specific defect, it is highly desirable to develop a fast and accurate method to identify the signal peptides. In this paper, a scaled window model is proposed. Based on such a model as well as Markov chain theory, a new algorithm is formulated for predicting the signal peptides. Test results for the 1939 secretory proteins and 1440 non-secretary proteins have indicated that the new algorithm is particularly successful in the overall success rate, and hence can serve as a complementary tool to the existing algorithms for signal peptide prediction.