Abstract
DNA microarray has been used widely to analyze functions of genes or drugs. The data shows genome-wide-transcriptome state of samples. Data-repository databases, such as Gene Expression Omnibus (GEO), provide a large number of experiment data. By comparing various gene expression profiles, we will be able to find biological connections between experiments, such as phytohormone and its inhibitors. However, microarray data is noisy snapshot. It is very difficult to compare biological similarity of gene expression across different laboratories, tissues. It is required to exclude noise to detect biological similarity between experiments.
To compare various publicly available microarray data, we applied network analysis using Co-module method in Arabidopsis. As a result of model data analysis, our approaches were successful to detect biological relationships between experiments from different laboratories. In addition, negative or partial connections were also detected. We constructed a novel Web-based database as an analysis tool of gene expression profiles.
