[BioC] Input to vsn or limma of genepix file missing 635 columns

[Wolfgang Huber]

Dear Dinos,

> Thanx for your reply.  So to make sure:
> 
> 1.  There is no read method in common packages (vsn, marray, affy,
> limma) that can read a single dye genepix file (Cy5/635 related columns
> missing) 

Just for the record:
No. See, for example, recent posts by Gordon and Eric:
https://stat.ethz.ch/pipermail/bioconductor/2008-May/022338.html
https://stat.ethz.ch/pipermail/bioconductor/2008-May/022310.html

> 2.  It is not recommended that I manually add those columns and fill
> them with 0s or N/A.

Correct. It is error-prone and difficult to reproduce.

> 3.  The best way to proceed is to use vsn2 on a matrix

Using vsn2 is one way, and I find it very useful in many cases. Whether 
it is the best (and indeed the definition of "best") depends on the 
quality of your data and your experimental question.

  Best wishes
	Wolfgang.
------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber

> Related to point 2, I actually tried that and was able to use marray
> read.GenePix which created an marrayRaw object.  It seemed alright but I
> could not reproduce any qc images.  I also could not convert it to an
> eset using convert even though there is a method that promises such a
> conversion.  Same happened when I created an maNorm object also.
> Finally ?vsn2 did not work on R2.7 and the latest vsn package version.
> I think the default path for the cfm file is incorrect (actually non
> existent).  However I did manually read the vignette/manual from the
> library/vsn/html directory.
> 
> Thanx again
> Dinos
> 
> -----Original Message-----
> From: Wolfgang Huber [mailto:huber at ebi.ac.uk] 
> Sent: Thursday, May 08, 2008 4:38 AM
> To: Liolios, Konstantinos [BSD] - MED
> Cc: bioconductor at stat.math.ethz.ch
> Subject: Re: [BioC] Input to vsn or limma of genepix file missing 635
> columns
> 
> Liolios, Konstantinos [BSD] - MED wrote:
>> Dear bioconductors,
>>
>> I have been given Exiqon miRCURY(tm) LNA Array (miRNA) expression gpr 
>> files in excel format.  They are single channel so they are actually 
>> missing all 635 related (mean/median/back/for etc.) columns.  I 
>> believe vsn is the way to normalize these data but I can not find 
>> anywhere how to import such a file.  Most tutorials use data that 
>> already come with the package or assume you have .cel files and an 
>> affymetrix annotation file.  I have used read.table which creates a 
>> data.frame but most normalization packages require you to have your 
>> data as an expression set.  I was wondering if somebody knows how to 
>> import single channel gpr files or how to manually create an 
>> expression set from a data.frame w/o an annotation file.
>>
>> Best regards
>> Dinos
>>
> 
> Dear Dinos,
> 
> there is a vsn2 method for "matrix", so please try converting your
> data.frame into a numeric matrix. Try
> 
> library("vsn")
> ? vsn2
> ? as.matrix
> 
> Also, the vignette "An introduction to Biobase and ExpressionSets",
> which you can access for example by typing
> 
>   openVignette()
> 
> contains a fairly good set of instructions for making an ExpressionSet. 
> Feedback on possible improvements to the documentation are always
> welcome.
> 
> Best wishes
>    Wolfgang
> 
> ------------------------------------------------------------------
> Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber
>