Thank you very much for all the answers. The data is ancient; those are the read-outs of spot intensities which were transferred to a CSV. As mentioned, the labeling is radioactive, so it is "single channel", and I just create the RG object manually from data frames read into R, and just create the design matrix with model.matrix() -- exactly the same procedure as the one I use for single channel Agilent. I will try to look into both, cellHTS and the nec() function. Best regards, January On Thu, Nov 11, 2010 at 11:45 PM, Gordon K Smyth <smyth at="" wehi.edu.au=""> wrote: > Dear January, > > As Wei says, the neqc() function in limma has the effect of subtracting the > mean intensity of the negative control or "background" spots from the other > spots, before going on to do other normalization. ?The nec() function gives > you a bit more control if you don't want to do quantile normalization. > ?These functions can operate on the objects you get from read.maimages(). > ?This is the way we'd recommend you to do it, although you could simply > background subtract without the normexp step. > > We could give more details in terms of code if you show us how you're > reading the data in and what sort of data object you're creating. ?For > example, is the data one channel or two channel? > > Best wishes > Gordon > >> Date: Thu, 11 Nov 2010 08:37:35 +1100 >> From: Wei Shi <shi at="" wehi.edu.au=""> >> To: January Weiner <january.weiner at="" mpiib-berlin.mpg.de=""> >> Cc: BioC <bioconductor at="" stat.math.ethz.ch=""> >> Subject: Re: [BioC] Background correction with just a few spots >> >> Dear January: >> >> ? ? ? ?The function neqc in limma package uses intensities from negative >> control probes to perform a normexp background correction, followed by >> quantile normalization and log2 transformation. For the details of this >> method, please see the paper: >> >> http://nar.oxfordjournals.org/content/early/2010/10/06/nar.gkq871.a bstract >> >> ? ? ? ?In brief, this method fits a normal+exponential convolution model >> to the data but use the negative control probe intensities to estimate the >> mean and standard deviation of background intensities. >> >> ? ? ? ?Let me know if you have any further questions. >> >> Cheers, >> Wei >> >> On Nov 10, 2010, at 7:56 PM, January Weiner wrote: >> >>> Dear all, >>> >>> I have a set of "strange" microarrays (nylon membrane / radioactive >>> labels). The raw data contains signals for the gene probes (a small >>> microbial genome) and for a number of probes which constitute the >>> background. There is no background signal directly in the data (like >>> in regular microarray chips), and I would like to subtract background >>> that is calculated from these few "background spots". Currently, I >>> just subtract the average of the background spots from all the other >>> spots. >>> >>> In limma, what would be the most appropriate way to do it? >>> >>> Cheers, >>> j. >>> >>> -- >>> -------- Dr. January Weiner 3 -------------------------------------- >>> Max Planck Institute for Infection Biology >>> Charit?platz 1 >>> D-10117 Berlin, Germany >>> Web ? : www.mpiib-berlin.mpg.de >>> Tel ? ? : +49-30-28460514 > > ______________________________________________________________________ > The information in this email is confidential and intended solely for the > addressee. > You must not disclose, forward, print or use it without the permission of > the sender. > ______________________________________________________________________ > -- -------- Dr. January Weiner 3 -------------------------------------- Max Planck Institute for Infection Biology Charit?platz 1 D-10117 Berlin, Germany Web?? : www.mpiib-berlin.mpg.de Tel? ?? : +49-30-28460514