Modeling Analysis Of Gene Differential Expression Based On Wavelet

Differentially expressed genes are composed of biological diversity,the same genes’ expression change with the external conditions,such as different tissues, different environments and different time points.Thus, differential expression genes ’ identifiction can help people understand the biological process and they are the foundation of the gene regulatory network’s building.Differential expressed genes can narrow scope of the study to improve the efficiency of building a network for the next biological, and can be used to analyze cancer research and drug targeting.Time-course expression data contains changes over time, it ’s the essential of building regulation network. Therefore, the study of differential gene expression based on the time is an important one to promote the work of bioinformatics and biology systems research information.In this paper, according to the current methods to identify differentially expressed genes, we analysis the disadvantages of various methods and problems needed to be solved. we proposed a model based on wavelet analysis to resolve the dissdvantages of Fourier Transform. The wavelet analysis has advanteges of solving the problems on time and frequency domain at the same time.we proposed a new cluster algorithms which can cluster fast and decide the centers by itselfs. At last, we use the GO analysis the differential expressed genes which clusterd together by the gostat toolbar.In this paper, we propose a model based on wavelet translationto which has strong function on local analysis identify differential expressed genes. we used the regression model with the wavelet analysis to overfit the time-course data.We used the wavelet analysis with multiscale to denoise the original data, then use the complex wavalet tree translation to get the wavelet coefficients on every scale. At last we calculate the similarity of every gene based on the wavelet coefficients of its.Cluster analysis divides the different genes into different classes, ensure the genes are similarity in the same cluster, and different from other classes. Thus we can get the unkonwn genes’ function according to the known genes in the same class. We present a fast automatic clustering algorithm in this paper, the ideas comes fro m a paper published on science 2014. The algorithm uses a Gaussian kernal function to calculation the genes density, and using the Particle Swarm Optimization to find the best threshold that decides the centers of t he class. This step decrease the time which used on analying the decision graph and reduce the influence of man-made on cluster.We composed the algorithm with the origianl on five datasets that are often used to research on cluster analysis, It demonstrate that the algorithm we proposed are feasibility and stable on cluster, it also has the same time complexity. At last,we design and develop a tool based on the algorithm used to analyze the data clustering and cisualization of results.