[BioC] Normalization for use of individual channels from 2-color arrays?

[Sean Davis]

On Tue, Sep 13, 2011 at 10:35 AM, Hari Easwaran <hariharan.pe at gmail.com> wrote:
> Dear BioC gurus,
>
> I am trying to use the individual channels from a 2-color array as a measure
> of gene expression level following the scripts in the Limma User Guide.
>
> The code the guide suggests is:
>
> MA <- normalizeBetweenArrays(MA, method="Aquantile")
>
> Throughout the guide, MA is an objected created by normalizing the RG values
> [for example using: MA <- normalizeWithinArrays(RG, method="loess")].
> So for the command above, are the A values obtained from a 1st step of
> normalization (like loess in this case) re-normalized using Aquantile?
>
> Since the idea of using Aquantile is so that the log ratios are not changed
> but the samples are adjusted to have the same distribution of A values, my
> concern is that by using a 1st step of normalization the log-ratios do not
> remain untouched. Does the guide mean to do the following:
>
> MA <- normalizeBetweenArrays(RG, method="Aquantile")
>
> where RG is the raw data.

Hi, Hari.

normalizeBetweenArrays is often used AFTER normalizeWithinArrays for
two-color arrays, so an MAList is often used as the object for
normalizeBetweenArrays.  Per the help page, the Aquantile method does
leave the M-values unchanged.

Sean