[BioC] affxparser fails to convert AGCC format cel file

[Steiner, Guido]

Henrik,
v1.10.1 works for me, too. Thanks for the fix!
I can convert the CELs to the old format now and then
use ReadAffy.

Concerning affyio, there is still a problem leading to a crash
on our system (IA64 Linux) when trying to read the Command Console
format directly. (Ben is looking into this.)

Best,
Guido

-----Original Message-----
From: henrik.bengtsson at gmail.com [mailto:henrik.bengtsson at gmail.com] On
Behalf Of Henrik Bengtsson
Sent: Wednesday, October 17, 2007 3:31 AM
To: bmb at bmbolstad.com
Cc: Steiner, Guido {PRBI~Basel}; bioconductor at stat.math.ethz.ch
Subject: Re: [BioC] affxparser fails to convert AGCC format cel file

Hi,

FYI, I've just committed a bug fix to affxparser (both release and
devel versions), but it will take a day or two before the builds
becomes available on all platforms.  Keep an eye out for affxparser
v1.10.1 on
http://bioconductor.org/packages/2.1/bioc/html/affxparser.html.

DETAILS: The previous code relied on the fact that there was a
(parent) header field 'affymetrix-dat-header' in the AGCC (Calvin) CEL
header.  However, in Guido's case the AGCC CEL files did not have such
a field but 'affymetrix-partial-dat-header'.  I've update the code to
also look for this and to some tricks to create a valid old-style DAT
header, which is required in the binary (v4) CEL headers.  If neither
exists, a more informative error message is now produced.

BTW (Benilton), convertCel() will no longer generate a warning if the
corresponding CDF file is not found.

Cheers

Henrik

On 10/11/07, bmb at bmbolstad.com <bmb at bmbolstad.com> wrote:
> Henrik,
>
> I can send you the file later if you want if you can't get it from
Guido
> (since I have received it myself), but you have correctly surmised
that
> there is a problem related to the non-existence of the DatHeader in
this
> particular CEL file. It is not how ever true to say that the DAT
header is
> not readily available in the AGCC CEL file, provision is made for
storing
> it in a parental header in the file format documentation (although
that is
> no guarantee that it will exist, the test files I had been using for
> affyio all had it existing). But I am sure you'd know better than I
how
> easy that is to access via the Fusion SDK.
>
> Interestingly enough I was able to successfully read this file just
fine
> with my installation using ReadAffy(). Only the affyio function
> read.celfile.header() with info="full" failed in my brief testing.
>
>
> Ben
>
>
>
>
> > Hi Guido,
> >
> > regarding affxparser: When writing a v4 CEL header, affxparser need
> > the DAT header, which is not readily available in AGCC CEL files so
> > affxparser will try to generate a valid one from what is available.
> > It seems like it fails in this case.  I cannot say more until I see
> > the CEL file.  Please let me (here or offline) know how I can
download
> > this CEL file (I cannot receive attachments > 19MB).
> >
> > /Henrik
> >
> > On 10/10/07, Steiner, Guido <guido.steiner at roche.com> wrote:
> >> Hi,
> >>
> >> I just tried affxparser on an example CEL file from the new
Affymetrix
> >>
> >> Command Console software.
> >>
> >>
> >>
> >> > library(affxparser)
> >>
> >> > convertCel("agcc.cel","gcos.cel")
> >>
> >> Error in sprintf("DatHeader=%s\n", datHeader) : zero-length
argument
> >>
> >>
> >>
> >> > createCel("test.cel",header=readCelHeader("agcc.cel"))
> >>
> >> Error in sprintf("DatHeader=%s\n", datHeader) : zero-length
argument
> >>
> >>
> >>
> >> So the error seems to happen at the createCell stage already.
> >>
> >> Reading the cel file with readCel does work!
> >>
> >> I have R 2.6.0/BioC 2.1 and affxparser version 1.10.0.
> >>
> >> Chip type is RAT_230 (I can provide the cel file if someone is
> >>
> >> interested).
> >>
> >>
> >>
> >> Any ideas how to solve this? (As an alternative - it would be fully
> >>
> >> sufficient for me to use readCel if there was a method to get the
> >>
> >> output into an AffyBatch object.)
> >>
> >>
> >>
> >> Regards,
> >>
> >> Guido
> >>
> >>
> >>
> >>
> >>
> >> --
> >> Guido Steiner, PhD
> >> F. Hoffmann-La Roche AG
> >> PRBI-M Scientific Methods and Algorithms
> >> Building 65/318B
> >> CH-4070 Basel
> >> Tel. +41 61 688 3329
> >>
> >> Confidentiality Note: This message is intended only for
> >> ...{{dropped:12}}
> >>
> >> _______________________________________________
> >> Bioconductor mailing list
> >> Bioconductor at stat.math.ethz.ch
> >> https://stat.ethz.ch/mailman/listinfo/bioconductor
> >> Search the archives:
> >> http://news.gmane.org/gmane.science.biology.informatics.conductor
> >>
> >
> > _______________________________________________
> > Bioconductor mailing list
> > Bioconductor at stat.math.ethz.ch
> > https://stat.ethz.ch/mailman/listinfo/bioconductor
> > Search the archives:
> > http://news.gmane.org/gmane.science.biology.informatics.conductor
> >
>
>
>