[BioC] gcrma and combineAffyBatch

[James W. MacDonald]

Hi Michael,

Michael Palumbo wrote:
> hello,
> 
> i'm new to bioconductor and processing array data, but i'm not new to R 
> - not an expert, either, let's say mildly competent...
> 
> (output of sessionInfo() is at bottom of this email.)
> 
> the short version of my question:
> i'm having trouble running gcrma() on an affy batch object returned from 
> combineAffyBatch(). the combined object is created by combining data 
> from two types of mouse chips: moe430a and mouse4302. gcrma() returns 
> this error:
> 
>  > gcrma.eset.A.V2 <- gcrma(comb.A.V2$dat)
> 
>  Adjusting for optical effect...Done.
> 
>  Computing affinitiesError in getCDF(cdfpackagename) : The current operation could not 
>  access the Bioconductor repository. Please check your internet connection, and report 
>  further problems to bioconductor at stat.math.ethz.ch
> 
> 
> the long version of my question, and more code:
> i believe the combining of chip data occurs without error. i put the 
> newly computed cdf in the global environment - i think this is where the 
> problem is - like it's not being recognized.

So there are two problems here. First, getCDF() looks for an _installed_ 
cdf package, rather than an environment, and if it doesn't find the 
installed package it tries to get one from BioC, which as you know is 
failing. I could envision some sort of hackery that could get around 
this problem. However...

The second problem is that you don't have a probe package that 
corresponds to the cdf you are using. Without the probe package you will 
not be able to proceed anyway, so the first point is sorta moot.

If you are really wedded to the idea of using gcrma(), I suppose you 
could get the probe_tab files for each of the chips you are using (from 
the Affy website), use awk or Perl or some such to come up with a 
combined probe_tab file that would correspond to the combined cdf that 
you are using, build the probe package, do some hack to getCDF() to find 
the cdfenv and then proceed happily.

Or else you could forget all that stuff, do RMA on your data, and 
proceed (slightly less) happily, but without spending all the time 
required to do the above.

Unless there is some obvious solution that escapes me. There are others 
here who use gcrma() all the time who may have already met and 
surmounted this problem. Although, as you noted, nobody is talking about 
it ;-D.

Sorry I can't be more helpful...

Best,

Jim


> 
> 
>>library(mouse4302cdf)  
> 
> 
>>library(moe430acdf)
> 
> 
>>comb.A.V2 <- combineAffyBatch(list(A,V2),c("moe430aprobe",
> 
>   "mouse4302probe"),newcdf="comb.430a.4302.cdf")
> 
>  package:moe430aprobe    moe430aprobe    
> 
>  package:mouse4302probe  mouse4302probe  
> 
>  245487 unique probes in common
> 
> # put new cdf in global environment
> 
> 
>>comb.430a.4302.cdf <- comb.A.V2$cdf
> 
> 
>>library(gcrma)
> 
> 
>   Loading required package: splines
> 
> 
>>gcrma.eset.A.V2.3 <- gcrma(comb.A.V2$dat)
> 
> 
>   Adjusting for optical effect...Done.
> 
>   Computing affinitiesError in getCDF(cdfpackagename) : The current operation could not 
>   access the Bioconductor repository. Please check your internet connection, and report 
>   further problems to bioconductor at stat.math.ethz.ch
> 
> ------
> when i check the cdf of the combined object, it looks correct to me:
> 
> 
> 
>>cdfName(comb.A.V2$dat)
> 
> 
> [1] "comb.430a.4302.cdf"
> 
> 
> --------------------
> 
>  > sessionInfo()
> R version 2.4.1 (2006-12-18)
> i386-pc-solaris2.10
> 
> locale:
> /en_US.ISO8859-1/C/C/en_US.ISO8859-1/en_US.ISO8859-1/C
> 
> attached base packages:
> [1] "splines"   "tools"     "stats"     "graphics"  "grDevices" "utils"   
> [7] "datasets"  "methods"   "base"    
> 
> other attached packages:
>  hgu95av2probe    hgu95av2cdf      hgu95acdf mouse4302probe   moe430aprobe
>          "1.0"        "1.4.3"        "1.4.1"       "1.14.0"       "1.14.0"
>          gcrma     moe430acdf   mouse4302cdf    matchprobes           affy
>        "2.6.0"       "1.14.0"       "1.14.0"        "1.6.0"       "1.12.2"
>         affyio        Biobase
>        "1.2.0"       "1.12.2"
> 
> i've searched the email list (what i think is) extensively and although 
> i've found other mentions of similar problems i haven't stumbled across 
> a solution that works for me. in particular, i tried a fairly different 
> approach that involved creating affinity info objects and that didn't 
> work either (i can supply the error of that attempt if necessary). the 
> method i tried is described here 
> <http://thread.gmane.org/gmane.science.biology.informatics.conductor/2500/focus=2747>.
> 
> thanks in advance for any and all help,
> mike
> 


-- 
James W. MacDonald, M.S.
Biostatistician
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109
734-647-5623


**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues.