Gene Prediction Algorithm Based On Wavelet Transform

With the rapid development of technology, there has been increasing focus onbioinformatics. And with the completion of the human genome project, the bioinformaticsturns a new chapter. A mount of DNA sequences from sequencing are need to analysis andprocess. Currently one of the commonly used prediction method is applied the method ofsignal processing to the biological characteristics of gene sequences, and predict the exonsof sequence according to the cycle characteristics of gene sequences. On the basis of formerstudies, this paper assured the independence of the model by wavelet transform, anddiscussed the best threshold selection of different species to improve the predictionprecision of the exons.This paper studies the gene exons prediction problem in the field of bioinformatics.According to the basic principles and the predicted model of DNA, mainly introduces thecommonly used prediction of spectrum analysis method in process: discrete Fouriertransform, short-time Fourier transform, Gabor transform and wavelet transform, andpresents the evolution process of several kinds of transformation methods and theiradvantages and disadvantages. At last, this paper gives the core ideas of gene prediction——gene prediction model based on wavelet transform. First, this paper introduces thebackground knowledge of bioinformatics, and the research status and significance of geneprediction. Then the paper introduces the basic principles of DNA sequencing, gives severalnumerical mapping methods of sequences, and explains the period three features of codingregions. Feature of three period as a basis classification for the paper’s research, whenexploring how to define a threshold for the classification, this paper gives two commonmethods of the feature extraction, domain and frequency methods. In the time domainfeature extraction, paper illustrates the characteristics and advantages of against notch filter,and presents a fast algorithm for power spectrum extraction: a fast transform algorithm forHartley based on time-domain extraction. In the frequency domain feature extraction, papergives three ways to calculate the power spectrum and noise ratio which based on threemapping modes. And then paper analyzes and compares the three methods which mentionsin the paper before. At last, this paper discusses the determination of threshold and analysisexperimental results. Take the mammals as an example, the certain threshold of humans andrats are presented. At the same time, some experiment results of gene prediction model based on wavelet transform are analyzed. This paper adopts ROC curve to evaluate the plan,and uses a large amount of data to support the ideas which claimed in the paper before.In one word, the method of wavelet transform is used to assured the independence ofmodel. And the experiment results prove that the model improve the efficiency of predictionprecision.