[BioC] affy chips' quality control question

[Na, Ren]

Hi,

 

First, I am very sorry that my attachments are still over size limit.

I couldn't find a way to make it smaller and readable at the 

same time.

I got 30 affy arrays (hgu133a) which were carried out over a period 

of two and half years. There are six types of samples, a,b,c,d,e,f. 

Each type has different number of biological replicates except type

f (no technical replicates). I am going to do pair wise comparisons 

among these types.

For checking array's quality, I read previous posts about this 

subject and compared my plots with PLM Image Gallery. But how 

to decide which array is acceptable or not acceptable is still

puzzling me. How much difference in these plots across arrays is 

considered as serious difference? And corresponding arrays should 

be excluded in down-stream data analysis? 

sample c2 is obviously different from other arrays in density plot,

NUSE plot, and it has large residuals in residual plot and large

average background. Maybe it's better to remove this array. But how

about the arrays with difference not as big as c2. Please see 

attachment files.

In file "residual.png",I included four arrays'residual images which

has bigger artifacts comparing with other arrays.

 

Any comments and suggestions would be greatly appreciated. 

 

Thanks!

 

Ren

-------------- next part --------------
A non-text attachment was scrubbed...
Name: hist_NUSE_RLE.png
Type: image/png
Size: 7358 bytes
Desc: hist_NUSE_RLE.png
Url : https://stat.ethz.ch/pipermail/bioconductor/attachments/20060203/27e1eed9/hist_NUSE_RLE.png
-------------- next part --------------
A non-text attachment was scrubbed...
Name: residual.png
Type: image/png
Size: 42943 bytes
Desc: residual.png
Url : https://stat.ethz.ch/pipermail/bioconductor/attachments/20060203/27e1eed9/residual.png