Please use this identifier to cite or link to this item:
Robust Dynamic Models Estimation of Microarray Gene Expression Data
|Authors: ||Meng-Lin Wu|
|Contributors: ||NTOU:Department of Communications Navigation and Control Engineering|
Microarray;Gene expression;Genetic algorithm;Occam filter;Singular value decomposition;Characteristic mode
|Issue Date: ||2011-07-04
cells reproduce by duplicating their contents and then dividing into two. The repetition of this process is called the cell-cycle, and is the fundamental means by which all living creatures propagate. On the other hand, abnormal cell divisions are responsible for many diseases, most notably cancer. Therefore studying cell-cycle control mechanisms and the factors essential for the process is important in order to aid in our understanding of cell replication, malignancy, and reproductive diseases that are associated with genomic instability and abnormal cell divisions. Recent breakthroughs in microarray technology have enabled biologists to measure the number of transcripts made from every gene in an organism’s DNA. This microarray technology allows an unprecedented look at the state of a cell at a particular time within the cell-cycle. Due to the importance of understanding the cell duplication process, studies of transcriptional regulation during the cell-cycle of yeast were among the first experiments to be carried out using microarray technology. First, since microarray data sets often contain missing values due to various reasons, e.g. insufficient resolution, image corruption, dust or scratches on the slides, experimental error during the laboratory process or even robotic methods can create missing values, genetic algorithm are proposed to estimate the missing values in the microarray data. Then, applications of Radial Basis Function Neural Networks (RBF NN) to the characteristic modes modeling problem of gene expression will be investigated. Extra time points will be searched by interpolation such that the modeling ability of RBF NN can be improved. Furthermore, reconstruction of the observed microarray data matrix with random variables of Gaussian density of zero mean and known variance is explored. And the expression profiles for the multiple microarray data with the concerned effective characteristic modes will be reconstructed. Finally, the Occam filter which employs lossy data compression to separate signal from noise based on SVD, will be used to estimate the number of characteristic modes for reconstructing the noise-free microarray data and also modeling the characteristic modes by fitting of a linear time-discrete dynamical system. We can describe the time evolution of expression values by using a time translational matrix to predict future expression values based on their expression values at some initial time. The yeast cell-cycle data set is illustrated to show the validation of the proposed modeling methods.
|Appears in Collections:||[通訊與導航工程學系] 博碩士論文|
Files in This Item:
All items in NTOUR are protected by copyright, with all rights reserved.