英文摘要 |
The data produced from DNA microarray are very suitable for massively parallel comparative analysis of gene behavior. Especially, by analyzing gene expression data from DNA microarrays, many types of knowledge may be discovered, including gene regulation network, genes involved in cell cycle, as well as relations between genes and genetic diseases. In this paper, we propose a novel method, which we call inverted clustering, for identifying gene pairs with negative regulation relationship by using temporal gene expression data. An integral part of our method is the use of data inversion to assess if the temporal expression profiles of two genes are opposite. If a gene is negatively regulated by another single gene, this pair of genes is very likely to have opposite temporal gene expression profiles. Self-organizing feature map (SOM) is an extensively used technique for cluster analysis. As a result of integrating SOM and the proposed data inversion method in analyzing temporal expression data of yeast genes during a diauxic shift, we identify 45 pairs of yeast genes with most opposite temporal expression profiles. Although the proposed data inversion method is very simple, it would make the extensively used cluster analysis technique an effective approach for the identification of gene pairs with negative regulation relationship. |