Hierarchical cluster analysis of self-organizing maps.

Using the cluster software,22 we performed hierarchical clustering of a self-organizing map with the genes judged to be significantly regulated by SAM, which were expressed in either group at a significant level (average difference, > 200), and for which the “fold change” ( = mean of ITD/mean of WT average difference) was more than 2 or less than 0.5. IL-3–stimulated samples that had been excluded from SAM analyses were included in this analysis. Values from the Affymetrix software were normalized and median centered before the self-organizing map (SOM) algorithm was applied followed by hierarchical clustering. Red areas depict high expression and green areas depict low expression of a given gene in a given sample. The average difference and fold change values for each gene are presented in the table.