Figure 1

The influence of data analysis algorithms on grouping of samples. Principle components analysis (PCA) and hierarchical cluster analysis of DNA copy numbers of normal genomic DNAs (C2, C4, C5 and C6) and neuroblastoma cell lines (SK-N-SH/G, SK-N-SH/L, SK-N-AS and IMR-32). log2 copy number estimates calculated by WPP (left) and RMA (right) were analyzed for relative distance by PCA (top) and by hierarchical clustering (bottom). RMA separates two closely related cell lines (SK-N-SH/G and SK-N-SH/L) in PCA (Fig. 1B) and clustering (Fig. 1D), while WPP groups them together (Fig. 1A and C), clearly separated from normal genomic DNAs.