Bioinformatics Study with an Implementation of Two Feature Extraction Algorithms for Protein Sequences

Abstract

Predicting subcellular localizations of proteins is related to multi-label learning. A serial of computational approaches have been developed. This study focuses on the extracting protein features. The feature vector influences the performance of a predicting algorithm significantly. In this paper, two feature extraction algorithms named composition-transition-distribution and class pattern frequency were introduced and implemented in Java, respectively. This program provided a friendly graphical user interface where users can get these two kinds of features easily and quickly. Moreover, the results can be saved into a specified file for later use. Finally, this program can be compressed into a single jar file and runs on a computer which installed the proper JRE. We hope that this program would give researchers some help in the future.