Extracting Features from Protein Sequences Using Chinese Segmentation Techniques for Subcellular Localization

This paper proposes a new method for extracting features from protein sequences to deal with the problem of protein subcellular localization. The idea behind the method arises from Chinese segmentation techniques. We regard the amino acid sequences as text and segment them into words in a non-overlapping way. The words are predefined in a dictionary, which includes valuable words according to some criteria. Every word in the dictionary will be assigned a weight, and a matching strategy called maximum weight product is adopted for segmentation. By recording word frequencies, a given sequence can be converted into a feature vector. To evaluate the effectiveness of the proposed feature extraction method, two different kinds of classifiers are used to predict protein subcellular locations. The experimental results show that our method is superior to existing approaches in classification accuracy and reduces the number of dimensions of feature space at the same time.