Power spectrum and dynamic time warping for DNA sequences classification

Abstract

Similarity and alignment and are often used to classify DNA sequences. We have developed a new classifier to classify DNA sequence. First, our approach is used to extract the features of DNA strands. Second, the goal of our approach is to classify DNA strands according to the similarity elaborated by the alignment. Frequently, the performance of the classification of DNA sequences depends on the method that allows to extract the characteristics and calculation of the genomic similarity. Particularly, our approach consists of three different methods for improving the classification of the DNA sequences. This paper presents a new approach of classification of DNA sequence based on dynamic time warping (DTW) method. First, the binary indicator is used to code each nucleotide and the power spectrum is used to extract the characteristics. Secondly, the DNA sequence similarity matrix is evaluated by the dynamic temporal Warping. Third, pairwise comparison is used to classify DNA strands. Our approach solves the complex problem of presentation and structure of different groups of organisms. The experimental results of our classifier obtained are compared with other approaches based on the alignment and similarity of the DNA sequences. These results showed that our approach outperformed other approaches in terms of classification and running time. Here is a summary of the main contributions of this article: (1) Convert nucleotides from DNA sequences by applying binary coding. (2) Using power spectrum our approach extracts the characteristics of DNA sequences. (3) Elaborate the similarity matrix of the DNA strand signal by the Dynamic Time Warping method. (4) Use pairwise comparison to classify DNA sequences. The approach developed is efficient to solve the problems of classification of DNA sequences.