ArrayPipe Online Documentation

[← previous] [up ↑] [next →]

Sample Analysis

Quality assessment

Before deciding on how to process the data it is useful to generate some plots and visualize the data. Change the action settings to the following (remove chip visualization of background, leave flagging of markers, change chip visualization foreground to MA plot and add signal box plot for each print tip group):

Clicking on the 'Do it!' button brings up MA plots and box plots per printTip (or subgrid):


(go to original page)

In this plot the log2 ratios are plotted against the log2 intensities. In general, one can assume that the majority of genes are not differentially expressed and should therefore have a ratio close to zero. This means that in an MA plot one would expect the data cloud to be aligned with and centered on the y=0 line. Deviations of this are often caused by dye-effects. The above plot shows clearly that some corrections are necessary or would you believe that the majority of the genes are up to 4-fold up-regulated (2 on log2 scale)?

The MA plot of the next slide looks pretty good:


(go to original page)

However, looking at the boxplots per subgrid for that slide we can observe quite some variation:


(go to original page)

On a typical array the spots are arranged randomly and one would expect that all subgrids show fairly similar distribution of expression values with the median centered around 0.

A good solution for correcting this kind of bias is the 'printTip loess' normalization procedure. And since the plots of the background intensities show quite some variation in background, it might be worth including background correction as well.

[← previous] [up ↑] [next →]



Home  

last modified $Date: 2007/08/27 14:02:03 $
for questions or remarks e-mail karsten_hokamp@sfu.ca.