Skip to main content
. 2008 Nov 28;9:505. doi: 10.1186/1471-2105-9-505

Figure 1.

Figure 1

Removal of nonlinear intensity effects by NVSA. (A) Simulated scaling factor as a function of intensity in the constructed simulation data series D1 (blue), D1.25 (green), and Dnl (red). Gene and random effects are not introduced so that the scaling factor line can be revealed. The log2 true intensities (x-axis) and the corresponding log2 (true intensities × scaling factors) are plotted. (B-C) Typical intensity scatter plots illustrating the nonlinearity before normalization (B) and removal of nonlinearity after (C) NVSA normalization. D1_g0 data (x-axis) is constructed using a linear scaling factor of 1 (blue in A) and zero gene effects plus random effects. Dnl_g10ud (y-axis) is simulated with the nonlinear scaling factors shown in A (red), 5% up-regulated plus 5% down- regulated genes (a total of 10% gene effects) in addition to random effects. Each blue dot represents a feature on the simulated microarray. The red line is 45-degree reference line.