Unsupervised hierarchical clustering of four biliary epithelial cells, one immortalized cholangiocyte cell line and nine CC cells. Unsupervised hierarchical clustering separated the samples into two main groups, normal biliary epithelial cells (NBE) isolated from mucosal slices of normal bile ducts and ex-vivo cultured as described in Methods, and cholangiocarcinoma cells (CCC). Data are in matrix format, with columns representing individual cell lines and rows representing each gene. Red, high expression; green, low expression; black, no significant change in expression level between the mean and sample. A hierarchical clustering algorithm was applied to all cells and genes using the 1 - Pearson correlation coefficient as a similarity measure. Raw data for a single array were summarized using Illumina BeadStudio v3.0 and output to the user was as a set of 43,148 values for each individual hybridization. We selected 828 unique genes with a two-fold or greater difference from the mean and P < 0.01 by t-test, for hierarchical clustering analysis. Specific gene clusters (Cluster 1 through Cluster III) were identified in the hierarchical cluster of the genes differentially expressed in CCC compared with NBE. CC, cholangiocarcinoma; IMC, immortalized cholangiocytes.