Condition trees generated by cluster analysis. A hierarchical cluster analysis (Euclidean distance clustering algorithm) was performed using PermutMatrix , which identified 139 genes whose expression differed significantly (≥ 2 fold change with respect to controls) in at least at one of the 2 disease stages analyzed (preclinical and clinical). Colors represent the level of expression. Sample information is listed across the top. The names of the known genes are indicated. Note the distinct patterns of gene expression in control, preclinical and clinical animals.