Abstract
Genome-wide expression profiles in the yeast Saccharomyces cerevisiae responding to environmental stresses were analyzed using microarray. Two yeast strains, a laboratory strain (FY834) and a brewing one (IFO2347) which is more osmotic tolerant than FY834 strain were analyzed and the differences between the two strains in response to several environmental stresses were elucidated. Based on the analyses of DNA microarray data, the information for breeding osmosis tolerance strain was extracted. To analyze microarray data, a Self-Organizing map (SOM) was introduced to cluster gene expression data. The range of experimental errors was determined by multiple measurements of a same sample, and the effect of expected error on the clustering results of the SOM was simulated. The clusters obtained by the SOM were further clustered until the influence of experimental errors on the clustering results falls to a certain criterion. A hierarchical clustering was adopted to cluster the SOM results, and we could obtain the clustering results which cannot be further subdivided in consideration with the experimental errors. By introducing some other information such as sequences of transcription factors, gene ontology data and metabolic pathways, we found that in the obtained clusters the frequency distributions of some factors, such as genes having specific transcription factors and genes categorized to specific functions, are significantly deviate from those computed by randomized data. Moreover we identified some clusters which represent significant up-regulation or down-regulation only in the specific conditions. For example, a gene cluster in which gene expressions were up-regulated only in IFO2347 strain was identified, and which represents the difference in osmotic tolerance between these strains.
