A Recognition Algorithm Of Gene Exon Based On Fourier Transform And Numerical Mapping

As the human genome work completed preliminarily, an emerging discipline has been born, bioinformatics, which is combined with the development of computer technology and biological technology disciplines. It is the cross subject including mathematics, statistics, computer and biomedicine, and research the acquisition of biological information, processing and interpretation and so on various aspects. Gene identification is one of the important research content and is of great significance.Exon identification problem is studied in this paper. We introduced the structure and identification method for the prokaryotes and eukaryotes, and analyzed the basic principle and processes of exon recognition. By studying every identification steps, the problem of exon identification is solved by using the corresponding method of signal processing. We compared the pros and cons of various existing of recognition algorithm, and the improved algorithm is proposed.Firstly, the period3of exons is discussed. By studying the Autocorrelation Fourier Transform to the effect of rendering for the period3of exon and drawing power spectrum graphics using Matlab software, we find that the autocorrelation Fourier transforms is able to increase value in the N/3power spectrum for exon with the period3. We adjusted sequence after the adjacent two or more than two exons together, and then drawed power spectrum diagram,so the sequence after adjustment is more the period3.Secondly, the mapping methods are classified on the basis of the rules of numerical mapping methods, and we discussed every numerical mapping method to performance influence of exon of the3-base periodicity, and the correlation of the mapping of Voss, Z-curve and regular tetrahedron. The fast algorithm under the Z-curve mapping is put forward, by mean of simulation proving the correctness and feasibility of the theoretical reasoning. Combining with the model of the best threshold value, we got two models of the threshold value after analysis and improvement. Through the comparison of three threshold values by experiment, we will know it is superior.Finally, we will identify exon by using the adaptive window translation method to calculate the signal-to-noise within the window. According to the threshold value to distinguish exons and introns, and then the feasibility and validity of the algorithm are judged based on sensitivity, specificity and distinctness as well as the comprehensive evaluation.