[BioC] Limma : Problem with lmFit() and block

[Gordon Smyth]

>Date: Tue, 19 Apr 2005 10:46:51 +0200
>From: Laetitia Marisa <Laetitia.Marisa at cgm.cnrs-gif.fr>
>Subject: [BioC] Limma : Problem with lmFit() and block
>To: bioconductor at stat.math.ethz.ch
>Message-ID: <4264C57B.8080403 at cgm.cnrs-gif.fr>
>Content-Type: text/plain; charset=ISO-8859-15; format=flowed
>
>Hello,
>
>When I use the block variable in the lmFit, data in the MArrayLM have
>different dimension :
>
>  > dim(analysis$fit)
>[1] 6936    3
>
>  > dim(analysis$fit$coef)
>[1] 6936    3
>
>  > dim(analysis$fit$genes)
>[1] 2312    6

The problem here is that fit$genes and fit$coef should always have the same 
number of rows.

The problem has occured because you have specified both 'ndups' and 'block' 
to lmFit() simultaneously, whereas lmFit() can only handle one of these at 
a time.

I will change lmFit() so that it generates an error when ndups>2 and 
'block' are set simultaneously.

Gordon

>Here is the command line I've  passed and a description of the data I've
>used :
>
>analysis$fit <- lmFit( msData, design=analysis$effects$Designmatrix ,
>block = msData$block , ndups=msData$printer$ndups,
>spacing=msData$printer$spacing,  weights=NULL )
>
>  > dim(msData)
>[1] 6936   15
>  > msData$block
>  [1] 1 1 2 3 3 4 4 5 5 6 6 7 7 8 8
>  > msData$printer$ndups
>[1] 3
>  > msData$printer$spacing
>[1] 1
>
>  > analysis$effects$Designmatrix
>          A  B  C
>Array1   1  0  0
>Array2  -1  0  0
>Array3   1  0  0
>Array4   1  0  0
>Array5  -1  0  0
>Array6   0  1  0
>Array7   0 -1  0
>Array8   0  1  0
>Array9   0 -1  0
>Array10  0  0  1
>Array11  0  0 -1
>Array12  0  0  1
>Array13  0  0 -1
>Array14  0  0  1
>Array15  0  0 -1
>
>Am I doing something wrong or it is a problem in the lmFit function?
>
>Thanks a lot.
>
>Laetitia Marisa.