[BioC] Limma. How can I deal with two sets with different printing.

[Gordon K Smyth]

> Date: Thu, 24 Feb 2005 20:35:06 +0100 (CET)
> From: M PEREZ <perezperezmm at yahoo.es>
> Subject: [BioC] Limma. How can I deal with two sets with different
> 	printing.
> To: Bioconductor Bioconductor <bioconductor at stat.math.ethz.ch>
>
> Hi list,
>
> I have five replicates comparing wt vs. mut.
>
> But 3 slides came from one printing and the other two came from other different printing (both
> from the same company and the most of the genes are the same in both sets). I am not sure how can
> I deal with them using Limma.
>
> I assume that it is not possible to perform a global print-tip loess normalization, so, Could be
> it possible to normalize the two set independently and after that joint both datasets in just one,
> and then apply Linfit and ebayes? . I appreciate any suggestion

This question has been asked a few times on this list, but not for a while.

Yes, you have to normalize the two sets independently and then merge them together.  If your two
sets contain exactly the same probes, just in a different order, then you can use the marge()
method provided in limma for MAList objects.  If not, then you have to match up the probes on the
two sets yourself, and then combine the re-ordered MALists together using cbind().  See earlier
postings from me where I suggest the use of the printorder() function to assist with the matching.
 (As far as I can see, it isn't possible to provided automatic merging functions which cover all
possible cases, hence you have to do the matching yourself.)

Gordon

> Thank you.
> Manuel