[BioC] c version of VSN ? and Wolfgang's answer

Wolfgang Huber huber at ebi.ac.uk
Tue May 15 22:01:55 CEST 2007


Dear Roger,

> But I still have the question about normalization,
> As for ENCODE tiling arrays which have about 383k probes, I have two
> biological replicates, which permitted me to perform BetweenArray
> normalization with vsn. The result is great.
> 
> Now, as for  whole genome tiling array set that has 30 arrays ( as you said
> different features but the same sample) I only have one biological
> replicate, therefore, I can only do WithinArray normalization.Which way you
> think it's a better solution, or the results are similar:
> 1. perform vsn for each array,
> 2. combine the 30 arrays together(as a large single array), then perform 
> vsn

Both of these will not work. vsn is for normalizing between multiple 
arrays of the same design, on which there are different (but related) 
samples.

I have no good idea how to normalize your data, maybe you should ask the 
people who designed the experiment.

Best wishes
   Wolfgang

------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber



More information about the Bioconductor mailing list