nodalpoint.org - "Odd" microarray normalization - Comments

R objects, and contacting authors

chris — Thu, 24 Jan 2008 10:27:34 -0500

I seem to recall that the limma/rma packages also have an option to return R and G values. Otherwise you can back-calculate from M and A.

M = log2(R) - log2(G)
A = 0.5*[log2(R)+log2(G)]

If only normalized data is available online, your best bet is to contact the authors and ask for raw intensities - there is no way of recovering them from processed data (at least, I don't think so).

They're cDNA data, according

lbbros — Thu, 24 Jan 2008 06:27:01 -0500

They're cDNA data, according to the paper (Zhao H et al., PloS Medicine 2005).

To tell you the truth, I'm not sure at all about consistency - but it's like a "test run" for a completely different approach I'm using: I'm just trying to see if such an approach is even feasible for the algorithm I'm going to run over the normalized data.

cDNA or oligo data ? Are you

Greg — Wed, 23 Jan 2008 11:42:14 -0500

cDNA or oligo data ? Are you sure the consistency of the data is good enough to be doing single channel normalization ?

"Odd" microarray normalization

lbbros — Tue, 22 Jan 2008 07:59:31 -0500

Hello.

I've been recently working with two color array data and to implement some procedures I'll need to use data from the two channels separately. Thus, I can't really use normalization methods that give a ratio or an average value (such as the ones included in the limma R package), I'll need to normalize the two channels separately and keep them separated. As I mostly use Affymetrix microarrays, I'm stuck, more or less.

Is anyone able to give a hint on how to proceed? Thanks.