A Study Of Gene Prediction Method Based On Digital Signal Processing

In recent years, Genome projects have given rise to an exponentially growingamount of genetic information. To analyze and deal with the rapid expansion of thedata and completely note on the gene structure of DNA sequence, it is a moreefficient method to reveal the connotation of these data obtaining more usefulinformation for the human. But the existing methods have many limitations, andforced us to consider to seek more effective methods. So it is a hot topic forbiologists, computer scientists and mathematicians to research.This paper research to the gene prediction of methods from the digital signalprocessing point of view. This method is using the typical characteristics of"theprotein coding region of three periodic". General Fourier spectrum encoding proteinDNA sequence has a peak at N/3frequencies and no obvious peak value in the noncoding region in the N/3frequency. This method first transform base sequence into asequence of characters, then the Fourier transform, the peak part of the coding region,otherwise non coding region. However, statistical analysis by Fourier transform forDNA sequence will bring noise inevitably, only according to the result by Fourieranalysis for DNA sequences, it is difficult to assess the probable genes of DNAsequence exactly. So this paper from the perspective of noise removal can improvethe prediction efficiency. Wavelet transform can wipe out high-frequency noisecaused by the removal of Fourier transform. But for the first exon is short, threeweeks is not obvious sequence and some low-frequency noise, you need to use theFIR filter.However, the original FIR filter design defects, this paper improves thedesign of FIR filter based on the original, and the wavelet transform and the FIRfilter is effective combination, formed the prediction method of a novel gene, and themethod is applied to the resistance gene prediction, experiments, this method isfeasible and efficiency higher. In addition, the new method is very easy to use,anddoes not require any prior knowledge of the genome sequence, and operation is veryfast.