Hello, I have been playing around with bioconductor for a few days now, trying to normalize data I have from genepix software. No one at my institution uses Bioconductor for microarray analysis (in fact, to my knowledge, they don't do any normalization beyond using a reference sample for one of the channels) so there is little help here. I have successfully been able to read in individual files and make nice graphs with them using marrayPlots, but ultimately I would like to be able to do loess, scale normalization between slides, and finally some sort of bayesian analysis for differential expression. Right now I'm stuck on reading in my data using marrayInput. But I don't even know that marrayInput is the best package--should I be using limma? I've read oh i don't know--3, 4 vignettes?-- that each have some help on the subject of reading in data, but they use a data set--"swirl"--am I mistaken here or is swirl composed of 4 sets of microarray files that *have already been read in*? The widget doesn't help. Should objects like swirl correspond to treatment groups? Where in marrayInfo does one put information about the target sample? How does that get related back to actual array data? There's a lot of documentation on what the classes are, but not on how to use them. I don't expect answers to these questions--I've scoured vignettes and message boards for this type of info. I just want to know, can anyone point me in the right direction? -Dennis

Hi Dennis, Put all you gpr files and the gal file in the same directory. Set your working directory to where your data is. The type: library(marrayTools) library(marrayPlots) data <- gpTools(raw=TRUE) data@maGb <- data@maRb <- matrix(0,0,0) datanorm <- maNorm(data) write.xls(maM(datanorm), file="results.xls") The above 6 lines will perfrom all diagnostic plots, do the normalization and output the results. If you don't need diagnostic plots, just try library(marrayTools) gpTools(bg=FALSE, quality=FALSE, plot=FALSE) and it will out put the normalized data for you in the same directory. Cheers Jean ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Jean Yee Hwa Yang jean@biostat.ucsf.edu Lung Biology Center, Tel: (415) 476-3368 University of California, Fax: (415) 476-6014 500 Parnassus Avenue, MU 420-W, San Francisco, CA 94143-0560 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ On Wed, 10 Mar 2004, Dennis Hazelett wrote: > Hello, > I have been playing around with bioconductor for a few days now, trying > to normalize data I have from genepix software. No one at my institution > uses Bioconductor for microarray analysis (in fact, to my knowledge, > they don't do any normalization beyond using a reference sample for one > of the channels) so there is little help here. I have successfully been > able to read in individual files and make nice graphs with them using > marrayPlots, but ultimately I would like to be able to do loess, scale > normalization between slides, and finally some sort of bayesian analysis > for differential expression. Right now I'm stuck on reading in my data > using marrayInput. But I don't even know that marrayInput is the best > package--should I be using limma? I've read oh i don't know--3, 4 > vignettes?-- that each have some help on the subject of reading in data, > but they use a data set--"swirl"--am I mistaken here or is swirl > composed of 4 sets of microarray files that *have already been read in*? > The widget doesn't help. Should objects like swirl correspond to > treatment groups? Where in marrayInfo does one put information about the > target sample? How does that get related back to actual array data? > There's a lot of documentation on what the classes are, but not on how > to use them. I don't expect answers to these questions--I've scoured > vignettes and message boards for this type of info. I just want to know, > can anyone point me in the right direction? > -Dennis > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >

On Wed, Mar 10, 2004 at 03:50:06PM -0800, Dennis Hazelett wrote: > Hello, > I have been playing around with bioconductor for a few days now, trying > to normalize data I have from genepix software. No one at my institution > uses Bioconductor for microarray analysis (in fact, to my knowledge, > they don't do any normalization beyond using a reference sample for one > of the channels) so there is little help here. I have successfully been > able to read in individual files and make nice graphs with them using > marrayPlots, but ultimately I would like to be able to do loess, scale > normalization between slides, and finally some sort of bayesian analysis > for differential expression. Right now I'm stuck on reading in my data limma and EBarrays are options for Bayesian analysis. Every package is supposed to have one or more vignettes that guide you through a basic analysis. You can find these using openVignette and you can explore them more thoroughly using the vExplorer function just go library(tkWidgets) vExplorer() > using marrayInput. But I don't even know that marrayInput is the best > package--should I be using limma? I've read oh i don't know--3, 4 > vignettes?-- that each have some help on the subject of reading in data, > but they use a data set--"swirl"--am I mistaken here or is swirl > composed of 4 sets of microarray files that *have already been read in*? > The widget doesn't help. Should objects like swirl correspond to which widget? > treatment groups? Where in marrayInfo does one put information about the > target sample? How does that get related back to actual array data? > There's a lot of documentation on what the classes are, but not on how > to use them. I don't expect answers to these questions--I've scoured > vignettes and message boards for this type of info. I just want to know, > can anyone point me in the right direction? > -Dennis You should not really need to know about the classes -> they should be relatively transparent at the user level. What you should find it easy to do is to read in your data and make a simple analysis. When I looked at the vignette for marrayInput I found that in Section 4 there were specific details for reading data in. The example might be a bit confusing - there are two parts to it, in one part the data have already been read in, as you identified, but the raw data files are also there and it is in Section 4 that the specific details are given. You can also find the raw data files by looking in the library directory for R - in marrayInput/data/ (and they are GenePix data). Hope that helps, Robert > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor -- +--------------------------------------------------------------------- ------+ | Robert Gentleman phone : (617) 632-5250 | | Associate Professor fax: (617) 632-2444 | | Department of Biostatistics office: M1B20 | | Harvard School of Public Health email: rgentlem@jimmy.harvard.edu | +--------------------------------------------------------------------- ------+

At 10:50 AM 11/03/2004, Dennis Hazelett wrote: >Hello, >I have been playing around with bioconductor for a few days now, trying to >normalize data I have from genepix software. No one at my institution uses >Bioconductor for microarray analysis (in fact, to my knowledge, they don't >do any normalization beyond using a reference sample for one of the >channels) so there is little help here. I have successfully been able to >read in individual files and make nice graphs with them using marrayPlots, >but ultimately I would like to be able to do loess, scale normalization >between slides, and finally some sort of bayesian analysis for >differential expression. Right now I'm stuck on reading in my data using >marrayInput. But I don't even know that marrayInput is the best >package--should I be using limma? For the processes you describe, you can use either limma or marrayInput/marrayNorm. It's just a matter of what style you prefer. Maybe have a look at one more document, the limma User's Guide. A little later down the track you might want to use limma for differential expression. At any stage, you can apply as.MAList() to your marray object and you're away. Gordon > I've read oh i don't know--3, 4 vignettes?-- that each have some help on > the subject of reading in data, but they use a data set--"swirl"--am I > mistaken here or is swirl composed of 4 sets of microarray files that > *have already been read in*? The widget doesn't help. Should objects like > swirl correspond to treatment groups? Where in marrayInfo does one put > information about the target sample? How does that get related back to > actual array data? There's a lot of documentation on what the classes > are, but not on how to use them. I don't expect answers to these > questions--I've scoured vignettes and message boards for this type of > info. I just want to know, can anyone point me in the right direction? >-Dennis

Section 4 of marrayInput.pdf shows you an example of how you would read raw data into the marray* packages. The functions you need are read.marrayRaw and read.Genepix. Reading the help on these really should make it easy to read data in. As for marray vs limma - for reading in and normalising they are both equally fantastic. If you are going to use limma for analysis though then it makes a bit more sense to read in in limma, simply for consistancy. -----Original Message----- From: Dennis Hazelett [mailto:hazelett@uoneuro.uoregon.edu] Sent: 10 March 2004 23:50 To: bioconductor@stat.math.ethz.ch Subject: [BioC] getting started Hello, I have been playing around with bioconductor for a few days now, trying to normalize data I have from genepix software. No one at my institution uses Bioconductor for microarray analysis (in fact, to my knowledge, they don't do any normalization beyond using a reference sample for one of the channels) so there is little help here. I have successfully been able to read in individual files and make nice graphs with them using marrayPlots, but ultimately I would like to be able to do loess, scale normalization between slides, and finally some sort of bayesian analysis for differential expression. Right now I'm stuck on reading in my data using marrayInput. But I don't even know that marrayInput is the best package--should I be using limma? I've read oh i don't know--3, 4 vignettes?-- that each have some help on the subject of reading in data, but they use a data set--"swirl"--am I mistaken here or is swirl composed of 4 sets of microarray files that *have already been read in*? The widget doesn't help. Should objects like swirl correspond to treatment groups? Where in marrayInfo does one put information about the target sample? How does that get related back to actual array data? There's a lot of documentation on what the classes are, but not on how to use them. I don't expect answers to these questions--I've scoured vignettes and message boards for this type of info. I just want to know, can anyone point me in the right direction? -Dennis _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor