| Budget Amount *help |
¥1,810,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥210,000)
Fiscal Year 2007: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2006: ¥900,000 (Direct Cost: ¥900,000)
|
|---|
| Research Abstract |
We performed a research to describe differences of shapes and works of cells by gene modules. First, as a data resource, we obtained expression data of 35,951 genes against 210 cells/tissues in total, from the collaborated researcher. And we obtained microscope images about 100 kinds of cells from the collaborator and others. Second, to measure the shapes of cells, we devised an image processing method that is robust against noise. Third, in order to extract combinatorial patterns of genes and cells exhaustively, we developed a fast biclustering method. With the two methods, we collected gene modules from gene expression data, while measuring parameters related to the contours of cells and nucleuses and the lining of cells. Analyzing the results, we obtained the whole set of gene modules that are commonly or reversely expressed among different cells. We found that this analysis allows us to perform digital differential display in more efficient way, for example, there was an extracted gene module that contained receptors, channels, transcription factors, etc., where genes were expressed reversely between brain and heart cells. Although it was not written in the proposal, to predict the shapes of cells from gene expression states, we performed the Support Vector Machine learning, and were able to predict the shapes of cells with 18 morphological parameters with 69.1 to 78.2% accuracies, against the test cells. We presented these results in two academic meetings, published three journal papers, and opened an Internet web service for retrieving and analyzing gene modules. Currently, we are writing a paper to summarize the whole study. Recognizing the values of the primary and the secondary datasets that we created in this project, our institute proceeded to support the development of the database from the fiscal year of 2008 to open it to public.
|
