Data storage for cluster analysis of microarray gene expression data

Microarray technology has now becoming a systematical way to study the expression level of thousands of genes over thousands of conditions. The large-scale, high-throughput experimental methods require analysis and information processing to match. Cluster analysis of gene expression data is one of the most important analysis steps in microarray technology. By using statistical algorithms, the purpose of cluster analysis is to group genes or samples together according to their similarities in gene expression profiles. In this thesis, we study the cluster analysis of microarray gene expression data, and analyze the data required for the clustering process. By realizing the MicroArray Gene Expression (MAGE) data standard, we design an object data model for cluster analysis of gene expression data, and construct a relational database schema. Furthermore, we integrate the database design to an existing open source microarray application--BASE (BioArray Software Environment). To validate the database, we modify a Java-based open source cluster analysis application--MeV (MultiExperiment Viewer). Therefore, we are able to take the gene expression data from BASE, to run the cluster application on MeV, and to store the generated clustering data back to the modified BASE database.