Figure 3

Trends from principal components. Based on 5,372 human microarray data Lukk et al.[1] showed that the first two principal components can be interpreted as "hematopoietic" and "malignancy" axes. In their study they used all 22,283 gene expression values in the principal component analysis. The two plots on the left side are reproductions of their results. Classifying the tissue of samples due to a "hematopoietic" direction (hematopoietic system, connective tissue, incompletely differentiated and other) a trend can be found along the horizontal axis (on top). Using another classification of samples (cell line, neoplasm, disease and normal) a "malignancy" trend can be recognized vertically (on bottom). Dots represent the samples colored according to classes determined by Lukk et al.[1]. Using the 250 most influential features filtered by leverage scores from CUR decomposition (with k = 5) very similar pattern was plotted (on right).