[BioC] [OT] compatible preprocessing for Affy and non-Affy

Adaikalavan Ramasamy ramasamy at cancer.org.uk
Wed May 23 05:25:52 CEST 2007


[[Apologies in advance since this may be an off topic subject but I have 
tried hard without finding any consensus.]]

Problem: I am trying to combine data from Affymetrix platform and 
non-Affymetrix platform (most of them two-color technologies) from the 
raw data (e.g. CEL, GPR). Therefore I need to find the choice of 
"compatible" preprocessing for Affy and non-Affy.

I found 3 studies trying to combine Affy with non-Affy. Irizarry et al 
(PMID:15846361) uses RMA for Affy and print-tip normalization with no 
background correction. Larkin et al (PMID:15846360) uses Li-Wong for 
Affy and LOWESS followed by z-scoring between array subgrids. I am 
unclear how Brammler et al (15846362) preprocessed their data (although 
one of the supplementary alludes to MAS 5.0 for Affymetrix).

Therefore, I have decided to use RMA for Affy and LOESS normalisation 
followed by z-scoring (i.e. zero mean, unit variance) the columns/arrays 
for non-Affy platforms.

Question: Can anyone kindly point me to any other papers that tries to 
address this question? Can anyone comment/correct me on my decision? Any 
other suggestions kindly welcome.

Thank you.

Regards, Adai



More information about the Bioconductor mailing list