[BioC] Converting data into MAlist to use in LIMMA

[J.delasHeras at ed.ac.uk]

Quoting Aubin-Horth Nadia <Nadia.Aubin-Horth at bio.ulaval.ca>:

> HI everybody
>
> We conducted a two-color microarray experiment using a 19 000-probe
> home made cDNA array. Our experiment contains 12 arrays. We use LIMMA
> to do all the normalization and model fitting and stats. Out of the 19
> 000 probes, several clones are part of the same contig, as annotated by
> TIGR. We decided to average the M values for these clones that
> correspond to a single contig to obtain a single M value for a given
> contig, for each array separately. We also wanted to remove probes that
> were called empty after sequencing (but they were already on the
> printed microarray). We exported the MAlist containing the normalised
> data (called "MAptip.nba.scale") and extracted the M data for each of
> the 12 slides in Python. We did the averaging and removing of "empty"
> spots and now have a new file with columns containing information on
> block, row, column, spot ID, annotation information for the contigs
> (and singletons) and then data for each slide in the following columns.
> Each row contains the averaged M values.
>
> We looked for a way to convert this file back into a MAlist so we can
> specify our design and do a fit. We read in the archives about a
> library called convert (which we did not find on CRAN) and info on how
> to transform data into an exprSet for affy data. Would someone be
> willing to help us with this task and give us pointers?
>
> Thank you very much
>
> Nadia Aubin-Horth

Hi Nadia,

it's actually quite simple, with Limma loaded just create a new MAlist  
like this:

newMA <- new("MAList")

and all you have to do is populate it with the appropriate components:

newMA$genes could be a matrix or a data frame containing your annotations
newMA$M is the matrix with your log2 ratios
newMA$A similarly containing the average expression values...
newMA$weights if you want to use weights...

I believe only the $M component is necessary. In fact, you don't need  
a MAList to use Limma, you could just feed a matrix of log2 values to  
lmFit(). But I'd use a fresh MAList with the M values and a $genes  
component. That way the annotations get passed on to the results,  
which I find convenient.

I hope this helps a bit.

Regards,

Jose

-- 
Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6513374
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR
UK
*********************************************
NEW EMAIL from July'09: nach.mcnach at gmail.com
*********************************************

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.