Figure 3

Clustering analyses using gene expression in PBMCs and the laboratory indices of disease progression. (a) Hierarchical clustering analysis using 998 nonredundant significant probe sets. The 998 nonredundant significant genes were normalized using Z-score calculation. Genes were clustered in Spotfire DecisionSite (Spotfire, Somerville, MA, USA). The correlation coefficient was used as distance metric and complete linkage was used as the clustering algorithm. (b) Hierarchical clustering of laboratory indices of disease progression. The laboratory indices for disease progression were used to cluster the samples. The measurements were normalized using the Z score across different animals and clustered in Spotfire DecisionSite, using the same algorithm as that for gene expression clustering, with correlation coefficient being used as distance metric and complete linkage as the clustering algorithm. The measurements are as follows: animal gross weight (weight), paw size (paw size), total white cell count (WBC), total lymphocyte count (LY), percentage lymphocyte of total WBCs (LY%), total monocyte count (MO), percentage monocyte count (MO%), total neutrophil count (NE), percentage neutrophil count (NE%), total eosinophil count (EO), percentage eosinophil count (EO%), total basophil count (BA), and percentage basophil count (BA%). Statistical tests were performed and the P value was attached for each measurement. Note that the phenotypic measurements separated the sample in a similar manner to the gene expression profiles. CIA, collagen-induced arthritis