Prediction of Subcellular Localization for Apoptosis Protein: Approached with a Novel Representation and Support Vector Machine

Apoptosis proteins play a crucial role in the development and home- ostasis of an organism. Obtaining information about subcellular location of these proteins is very important to understand the mechanism of programmed cell death. In this paper, based on the hydropathy characteristics, we introduce the frequency of 2-blocks and pK value of the α-NH + group of 2-blocks. By using the new representation for apoptosis protein sequence and support vector ma- chine, we predict subcellular location of 317 apoptosis proteins in jackknife test. The overall prediction accuracy is 91.80% which is higher than other existing algorithms. Furthermore, another dataset containing 98 apoptosis proteins is ex- amined in the same method. The overall predicted successful rate is 94.85%. The promising results indicate that our method may play a complementary role for predicting subcellular location of apoptosis protein.