Abstract—The type III secretion system (T3SS) is a complex structure which allows gram-negative pathogens to destroy eukaryotic cell biology by injecting virulence factors directly into the host cell cytoplasm. Composed of around 30 proteins, T3SS is among the most complex secretion systems identified in Gram-negative bacteria. Since type III secreted effectors (T3SEs) are essential for the pathogenicity, identification of T3SEs is one of the core problems in computational biology. This paper puts forward a new method for the prediction of T3SEs. The method is a sequence-based approach which can extract useful features from amino acid sequences. By calculating the frequency of the features from different segments of protein sequences, the data set is represented by the feature vectors and classified by Support Vector Machine (SVM). The experimental results show superiority over other available approaches on classification accuracy.

Index Terms—Type III secreted effector prediction, sequence-based approach, feature extraction, type III secretion, word segmentation, hybrid feature system.

Tingting Sui is with the Institute of Computer Application and Technology, Shanghai Maritime University, CO 201306 CHN (e-mail: suisui61@163.com).
Yang Yang and Xiaofeng Wang are with the Department of Computer Science and Engineering of Shanghai Maritime University, CO 201306 CHN (e-mail: yangy09@gmail.com, xfwang@shmtu.edu.cn).

 

Cite:Tingting Sui, Yang Yang, and Xiaofeng Wang, "Sequence-Based Feature Extraction for Type III Effector Prediction," International Journal of Bioscience, Biochemistry and Bioinformatics vol. 3, no. 3, pp. 246-251, 2013.