Unsupervised hierarchical clustering: Panel A.

<p>Heat map based on 8,210 genes of 71 liver samples (HCV-related HCC, non-HCC counterpart, metastases, non cancerous counterpart and controls from healthy donors). Genes were filtered according to the following criteria: presence in 80% of all experiments, a >3 fold change ratio in at least one experiment. Red indicates over-expression; green indicates under-expression; black indicates unchanged expression; gray indicates no detection of expression (intensity of both Cy3 and Cy5 below the cutoff value). Each row represents a single gene; each column represents a single sample. The dendrogram at the top of the matrix indicates the degree of similarity between samples. <b>Panel B</b> Principal Component Analysis (PCA) 3D view for gene expression profiles from 71 liver samples. The PCA is based on log2 ratios and the expression profiles are across all the 36,000 genes in the microarrays. The green, blue, red, purple, and orange dots indicate metastasis, HCV-related HCC, HCV related non HCC, non metastatic counterpart, HCV-negative normal control and samples, respectively.