Prediction of protein subcellular localization based on multilayer sparse coding.

Author: Xingjian CHEN ¹ ; Xuejiao HU ¹ ; Wei XUE ¹
Author Information

1. School of Information Science and Technology, Nanjing Agricultural University, Nanjing 210095, Jiangsu, China.
Publication Type:Journal Article
Keywords: amino acid composition; multilayer pooling; sparse coding; subcellular localization prediction; support vector machine
MeSH: Algorithms; Amino Acid Sequence; Computational Biology; Protein Transport; Proteins; Subcellular Fractions; Support Vector Machine
From: Chinese Journal of Biotechnology 2019;35(4):687-696
CountryChina
Language:Chinese
Abstract: In order to provide a theoretical basis for better understanding the function and properties of proteins, we proposed a simple and effective feature extraction method for protein sequences to determine the subcellular localization of proteins. First, we introduced sparse coding combined with the information of amino acid composition to extract the feature values of protein sequences. Then the multilayer pooling integration was performed according to different sizes of dictionaries. Finally, the extracted feature values were sent into the support vector machine to test the effectiveness of our model. The success rates in data set ZD98, CH317 and Gram1253 were 95.9%, 93.4% and 94.7%, respectively as verified by the Jackknife test. Experiments showed that our method based on multilayer sparse coding can remarkably improve the accuracy of the prediction of protein subcellular localization.