The study of threshold determination of gene identification and its improvement algorithms

Abstract

Abstract. The problem of gene recognition based on the ratio of power spectrum, SNR, and Gabor transform and its implementation of the calculation were discussed. The optimal threshold could guarantee to identify the DNA sequences with the signal-to-noise ratio. It summarized three kinds of traditional ways to determine the threshold, and advanced the optimum entitled method showing the disparate degrees of highlight and the discrimination rate method of the exons or introns as far as possible to improve the rate of their accuracy. To evaluate different determination methods of threshold by using the calculation results of four kinds of DNA sequence. In order to ensure the analysis of DNA sequence more accurate, it adopted and improved gene identification method of Fourier transformation in a short time which is based on Gabor transformation. By using of the ergodic theory, the fixed percentage of the sequence length of exons in DNA has been improved to be the dynamic percentages which focus on different gene types. The exons of the DNA sequence which have been already discovered were identified by using the improved algorithm. With comparison of the results and the actual endpoint of exons, it confirmed that the improved algorithm can figure out the endpoint of the exons more accurate.