Skip to main content
Figure 4 | BMC Bioinformatics

Figure 4

From: Iterative rank-order normalization of gene expression microarray data

Figure 4

Principle component analysis. Lung adenocarcinoma derived samples (triangles) separate from small-cell derived samples (circles). Replicate outliers are designated by their respective unfilled shapes. (A) IRON normalized; adenocarcinoma samples cleanly separate from small-cell samples along the first principle component. Biological separation is captured almost entirely by the first principle component, with negligible separation along the second component. Outlier replicate samples group with their respective cohort. (B) dChip normalized; results are similar to those of IRON, but separation of the two groups is slightly less clean, and the small-cell outlier is on the adenocarcinoma side of the first principle component. (C) MAS5 normalized; adenocarcinoma and small-cell samples cleanly separate from one another, together with their respective outlier samples. Two distinct sub-clusters are evident in both the adenocarcinoma and small-cell derived samples, which are not observed in the results from other methods. (D) RMA normalized; adenocarcinoma and small-cell samples do not separate well along the first component, requiring a combination of both the first and second components to achieve separation. Replicate outlier samples group with themselves, rather than with their respective biological cohorts.

Back to article page