[BioC] limma normalization

[Gordon Smyth]

At 11:26 PM 18/09/2003, Jason Skelton wrote:
>Hi
>
>trying to use limma with my own genepix data
>have a problem when I try to use the following commands:
>
> > dictyMA <- normalizeWithinArrays(DictylimmaRG, layout=dictylimmalayout, 
> method="printtiploess")
>Error in residuals(loess(y ~ x, weights = w, span = span, na.action = 
>na.exclude,  :
>    couldn't find function "loess"
>
> > dictyMA <- normalizeWithinArrays(DictylimmaRG, dictylimmalayout)
>Error in residuals(loess(y ~ x, weights = w, span = span, na.action = 
>na.exclude,  :
>    couldn't find function "loess"
>
>Any ideas ?

You need to type

library(modreg)

By the way, the fact that you are having this problem shows that you are 
using an old version of R and a version of limma somewhere between the last 
official release and the current development version. Updating either R or 
limma to the latest version will solve the problem permanently.

>Also for the design matrix
>e.g design <- c(-1,1,-1,1)
>I know the negative numbers are dye swaps but does it assume your normal 
>dye orientation is a certain way around e.g green/red OR red/green for the 
>positive numbers ?
>
>and therefore the  negative numbers(dye swap) is just the reverse or does 
>it actually matter atall ? (hope that makes sense)
>presuming of course that -1, 1 correspond to the order in which you have 
>read them into limma in the first place ?

It doesn't matter which way around you do it as long as you know and 
interpret the results the right way. I always do it so that upregulation in 
the red channel corresponds to positive log-ratios.

Gordon

>thanks
>
>Jason
>
>--
>--------------------------------
>Jason Skelton
>Pathogen Microarrays
>Wellcome Trust Sanger Institute
>Hinxton
>Cambridge
>CB10 1SA
>
>Tel +44(0)1223 834244 Ext 7123
>Fax +44(0)1223 494919