[BioC] RE: Dye Swap Experiments

[Gordon Smyth]

Dear Mike,

At 09:05 PM 13/06/2003, michael watson (IAH-C) wrote:
>Hi Gordon
>
>I have finally found some time to try out limma and have come across some 
>problems.
>
>I am using the following:
>
>Package: limma
>Version: 1.0.3
>Date: 2003/05/04
>Title: Linear Models for Microarray Data
>Author: Gordon Smyth <smyth at wehi.edu.au>
>Maintainer: Gordon Smyth <smyth at wehi.edu.au>
>Depends: methods, modreg, MASS, splines, nls, nlme, R (>= 1.6.1)
>Description:  Data analysis, linear models and differential expression for 
>spotted and oligonucleotide microarrays.
>License: GPL version 2 or newer
>URL: http://bioinf.wehi.edu.au/limma/
>
>
>I am having trouble calculating MA values.  Having read in some genepix 
>data using
>
>RG <- read.maimages(..., "genepix")
>
>I then got the following:
>
>MA2 <- normalizeWithinArrays(RG,lout,method="loess")
>Error in residuals(loess(y ~ x, weights = w, span = 0.75 * span, na.action 
>= na.exclude,  :
>         couldn't find function "loess"

Sorry, my mistake. The 'loess' function is part of the 'modreg' package 
which you don't have loaded. You just need to type

library(modreg)

at the R prompt and then the command that you give above will work fine. I 
thought that I had arranged for 'modreg' to be loaded for you automatically 
if needed, but apparently not. Note that this particular problem will 
disappear if you upgrade to R 1.7.0 or later because 'modreg' is now loaded 
by default as part of R.

>also, the following commands worked but produced THE SAME MA values:
>
>MA <-  normalizeWithinArrays(RG,lout,method="none")
>MA1 <-  normalizeWithinArrays(RG,lout,method="lowess")
>
>I suspected that method was defaulting to some default method so I tried:
>
>MA <-  normalizeWithinArrays(RG,lout,method="crap")
>
>and again got the same MA values.

You are quite right, if the method requested is not even a partial match 
for any of the possibilities, then no normalization is done.

>  Shouldn't there be a warning when the method is not recognised??

Yes, perhaps there should be.

>   Also, the documentation says method="loess" should work, but i just get 
> the above error message
>
>Otherwise this was looking like quite a nice package... :-D

Thanks for your comments.

>Thanks
>Mick

Note that 'limma' is up to version 1.1.0 now.

Gordon
---------------------------------------------------------------------------------------
Dr Gordon K Smyth, Senior Research Scientist, Bioinformatics,
Walter and Eliza Hall Institute of Medical Research,
1G Royal Parade, Parkville, Vic 3050, Australia
Tel: (03) 9345 2326, Fax (03) 9347 0852,
Email: smyth at wehi.edu.au, www: http://www.statsci.org