Fig. 16.

Comparing normalization method effect on significance designation using parallel coordinate plots. Subplot a shows parallel coordinate plots of the DEGs from liver and kidney technical replicates [12] after library size scale normalization. The division of DEGs between the two groups was rather disparate, with ∼78% of the DEGs being kidney-specific and only ∼22% of the DEGs being liver-specific. Also of note, while the parallel coordinate patterns of the liver-specific DEGs appear as expected, the patterns of the kidney-specific DEGs seem to show comparatively larger variability between the replicates. Subplot b shows parallel coordinate plots of the DEGs from liver and kidney technical replicates after TMM normalization. The division of DEGs between the two groups is more balanced than in Subplot a, with ∼53% of the DEGs being kidney-specific and ∼47% of the DEGs being liver-specific. Additionally, the parallel coordinate patterns of the kidney DEGs is vastly improved. However, the parallel coordinate patterns of the liver DEGs is slightly more messy looking. As a result, we investigate the effects of normalization on this data more thoroughly