Logo image
微陣列之影像掃瞄系統與群集分析之研究
Thesis

微陣列之影像掃瞄系統與群集分析之研究

陳彥良
Masters, National Tsing Hua University
1999

Abstract

微陣列微排玻片掃瞄系統群集分析樹狀群集主分量分析 cDNA microarrayscannerscan analyzecluster analysishierarchical clusteringpartitional clusteringK-means clusteringprincipal component analysis
DNA microarrays, microscopic arrays of large sets of DNA sequences immobilized on slides, are powerful tools in identifying or quantifying many specific DNA sequences in complex nucleic acid samples. DNA microarrays have been used in genetic mapping studies, mutational analysis and in genome wide monitoring of gene expression, and will become standard tools in research and clinical applications.Making and using printed DNA microarrays requires two pieces of hardware, named arrayer and scanner. The mRNA samples turn into cDNA targets, which were labeled fluorescence dyes simultaneously by reverse transcription. The different DNA probes printed previously on slides were hybridized with the cDNA. After cleaning the non-hybridized cDNA, the scanning results were analyzed by computer. In our scanning system, the limited concentration of fluorescent detection can achieve M, and it will suffice the microarray experiments.One of the advantages of microarray is quantifying the thousands of gene expressions simultaneously. The relation between different moments or conditions with the strength of gene expressions will be obtained by observing these experiments. Grouping the similar variation genes together will presume the function of the unknown genes by some genes we have known, named cluster analysis. The way of calculating distance between any two genes will be decided first. The computer simulations show the results of the hierarchical clustering and the K-means partitional clustering. We then design K-hierarchical clustering by acquiring their strong points. Besides, the principal component analysis is used in these simulations to reduce the errors of the hierarchical clustering and compare with the different clustering results.

Metrics

1 Record Views

Details

Logo image