## BDNF dataset

Variance stabilization was performed for each dye swap (two experiments) separately. Apart from ANOVA, all normalizations were used separately for each experiment. After normalization, the sample repetitions were averaged. Thus, we ended up with two datasets corresponding to the two RNA

samples (undifferentiated and differentiated cells) which have to be compared. As graphical display we used the representation of log product versus log ratio. This is shown in Figure 17.2. The dye-swap experimental data (6) shows a high percentage of low-expressed genes after background correction. Variance-stabilizing normalization leads to a very good correction of the small-intensity values while other methods cannot correct for this effect and still show - in logarithmic scale - a highly scattered plot in the low-intensities range. Due to the fact that for ZScore scaling the values are corrected for mean and divided by the standard deviation, this scattering effect is strengthened.

Figure 17.2.

Normalization of a dye-swap experiment. Each subplot displays the scatterplot of log-product (x-axis) versus the log ratio (y-axis) of the mean of one dye-swap repetition. The subplot in the upper left displays the mean values of background-corrected raw intensities without any normalization. The other subplots show scatterplots after application of global mean scaling, global median scaling, shorth scaling, Zscore normalization, global linear regression, global quadratic regression, loess regression, ANOVA normalization, variance stabilization, quantile normalization and qspline normalization.

### Figure 17.2.

Normalization of a dye-swap experiment. Each subplot displays the scatterplot of log-product (x-axis) versus the log ratio (y-axis) of the mean of one dye-swap repetition. The subplot in the upper left displays the mean values of background-corrected raw intensities without any normalization. The other subplots show scatterplots after application of global mean scaling, global median scaling, shorth scaling, Zscore normalization, global linear regression, global quadratic regression, loess regression, ANOVA normalization, variance stabilization, quantile normalization and qspline normalization.

## Post a comment