Accurate prediction of Gram-negative bacterial secreted protein types by fusing multiple statistical features from PSI-BLAST profile

Abstract

Gram-negative bacterial secreted proteins play different roles in invaded eukaryotic cells and cause various diseases. Prediction of Gram-negative bacterial secreted protein types is a meaningful and challenging task. In this paper, we develop a multiple statistical features extraction model based on the dipeptide composition (DPC) descriptor and the detrended moving-average auto-cross-correlation analysis (DMACA) descriptor by PSI-BLAST profile. A 610-dimensional feature vector was constructed on the training set, and the feature extraction model was denoted DPC-DMACA-PSSM. A support vector machine was then selected as a classifier, and the bias-free jackknife test method was used for evaluating the accuracy. Our predictor achieves favourable performance for overall accuracy on the test set and also outperforms the other published approaches. The results show that our approach offers a reliable tool for the identification of Gram-negative bacterial secreted protein types.

Keywords:

• Secreted protein
• position-specific scoring matrix
• dipeptide composition
• detrended moving-average auto-cross-correlation
• support vector machine

Acknowledgements

The authors would like to thank the anonymous reviewers and editor for their helpful comments on our manuscript.