[BioC] duplicate correlation on Agilent 4x44 arrays

[Gordon K Smyth]

Dear Mitch,

The data.frame RG$genes must have the same number of rows as the intensity data RG$R, RG$G etc
because it is intended to provide probe annotation information corresponding to each row of the
data.  Your GAL file doesn't satisfy this requirement, because it has massively more rows than the
data.

I am not familiar with the 4 x 44 technology, so I don't know what format the data files are
written to.  From the information you give, I am guessing that read.maimages() has read the data
from only one of the 4 blocks of data, or else that the 4 blocks of the 4x44 format have been read
into separate columns of the data.  I am guessing that the GAL file has the probes for all four
blocks.

Furthermore, some of the spots, perhaps empty probes, have been ommitted from your data files. 
This seems to have been done in an uneven way, because your data doesn't have an even number of
rows.  Considering this, you can hardly expect to use your data with duplicateCorrelation().

If you have hybridised different RNA samples to the four different blocks (you don't say), then
read.maimages() looks correct and your GAL file is incorrect.  I'm only guessing because I'm not
familiar with the 4 x 44 format.  You need to do some trouble-shooting of these issues at your
end.

Best wishes
Gordon

On Tue, April 10, 2007 10:07 pm, Mitch Levesque wrote:
> Gordon,
>
> Thanks for the reply. I am not using any particular instruction set, just
> what I have put together from the User Guide.
>
> You were right about the file dimensions, they are different:
>
>> dim(RG)
> [1] 44407     4
>> gal <- readGAL()
>> dim(gal)
> [1] 180880     10
>
> Is it possible to read the duplicate positions directly off of the gal file?
> I tried:
>
> layout <- getLayout(gal, guessdups=TRUE)
>
> and I get the following:
>
> $ngrid.r
> [1] 1
>
> $ngrid.c
> [1] 4
>
> $nspot.r
> [1] 170
>
> $nspot.c
> [1] 266
>
> $ndups
> [1] 8
>
> $spacing
> [1] NA
>
> attr(,"class")
> [1] "PrintLayout"
>
>
> I haven't tried without the normexp, but I will test it. Thanks again.
>
> Mitch
>
>
>
> -----Original Message-----
> From: Gordon Smyth [mailto:smyth at wehi.EDU.AU]
> Sent: Tuesday, April 10, 2007 1:03 PM
> To: Mitch Levesque
> Cc: bioconductor at stat.math.ethz.ch
> Subject: [BioC] duplicate correlation on Agilent 4x44 arrays
>
> Dear Mitch,
>
> You don't say what instructions you are trying to follow here. I
> think you may be trying to use code which was intended for other data
> sets. I suspect that there may be more than one problem.
>
> Firstly, why do you need to use readGAL()? This is only needed with
> SPOT data. Your RG object from read.maimages() will already contain
> annotation information from the Agilent output files. Look at
>
>     names(RG$genes)
>
> to see what you have.
>
> Secondly, does your GAL file match your data files? Type
>
>     dim(RG)
>
> and
>
>     gal <- readGAL()
>     dim(gal)
>
> Do the row numbers agree? I am guessing they may have different
> numbers of rows.
>
> BTW, do you need to use "normexp"? I've found the AgilentFE
> background estimator is already pretty good, and doesn't produce
> negative intensities anyway.
>
> Best wishes
> Gordon
>
>>Date: Mon, 9 Apr 2007 12:21:57 +0200
>>From: "Mitch Levesque" <Mitch.Levesque at tuebingen.mpg.de>
>>Subject: [BioC] duplicate correlation on Agilent 4x44 arrays
>>To: <bioconductor at stat.math.ethz.ch>
>>
>>Hi Bioconductors,
>>
>>I am using R 2.4.1 and limma to analyze the new Agilent 4x44 array design
>>and am having trouble with the duplicate correlation function using the
>>following script:
>>
>>
>>library(limma)
>>targets <- readTargets("Targets.txt")
>>RG <- read.maimages(targets$FileName, source="agilent")
>>RG$genes<-readGAL()
>>RG$printer<-getLayout(RG$genes)
>>RG <- backgroundCorrect(RG, method="normexp", offset=50)
>>MA <- normalizeWithinArrays(RG, method="loess")
>>MA <- MA[order(RG$genes[,"ID"]),]
>>
>>I get the following error:
>>
>>Error in `[.MAList`(MA, order(RG$genes[, "ID"]), ) :
>>        subscript out of bounds
>>
>>I would like to treat the duplicate probes on each array as a technical
>>replicate, but since the spacing is not consistent for each gene, I must
>>first order the list by reference number. Are there any suggestions about
>>how I may do this?
>>
>>Mitch
>
>
>
>
>