[BioC] marray intensity normalization?

Gordon Smyth smyth at wehi.edu.au
Fri Jun 17 12:37:06 CEST 2005

>Date: Thu, 16 Jun 2005 14:29:59 -0600
>From: Jake Michaelson <jjmichael at comcast.net>
>Subject: [BioC] marray intensity normalization?
>To: bioconductor at stat.math.ethz.ch
>Hi all,
>I'm a little more familiar with single-channel chips (Affy) than I am
>with spotted two-color arrays, and I'm now starting to get involved
>with the analysis of some two-color arrays.
>I was wondering (thinking of this from the single-channel mentality) if
>there's a way to use marray (or another package) to get normalized
>intensities of each channel, rather than log-ratios; this would
>essentially treat each channel like an individual single-channel chip.
>I'd like to then convert these normalized intensities into an exprSet,
>with two intensities per chip.  In other words, if I had 12 two-color
>chips, I'd like to end up with a 24-column exprSet.

You might read Section 24 "Between-Array Normalization of Two-Color Arrays" 
of the limma User's Guide, and investigate the 'convert' package which does 
exactly what you ask for.

>Somebody please stop me in my tracks if I'm heading into dangerous
>waters by using a single-channel mentality with two-color chips!

You are heading in very dangerous waters indeed and, almost certainly, you 
don't really want to do it. If you change your technology, you need to 
change you thinking as well. The red-green pairing is the whole point of 
the two-colour technology and you cannot ignore it without drastic loss of 



More information about the Bioconductor mailing list