Figure 3
From: Use of normalization methods for analysis of microarrays containing a high degree of gene effects
Use of NVSA for improving normalization accuracy. The NVSA variable selection function (denoted as nv) was used prior to quantile, lowess, or xcorr normalization analysis of Dnl data series. (A) The percentage of variant genes filtered out by the NVSA variable selection procedure increases along with the percentage of gene effects. (B) Normalization accuracy measured by the mean relative error versus the total percentage of gene effects represented by DEG. Up, down, up/down indicate the gene effects as up-, down-, and an equal mixture of both, respectively. Errors that are greater than 20% are not shown to allow a better view of details in the low error region.