read.celfile (from package affy version 1.1.0) keeps hanging with cel files exported from dChip version 1.2. I am not sure how standards- conform the dChip-export is, but in any case it would seem useful for read.celfile to produce an intelligible error message rather than just hanging. I managed to work around this by commenting the lines ff@masks <- .Call("getIndexExtraFromCEL", ... ff@outliers <- .Call("getIndexExtraFromCEL", ... Wolfgang. Dr. Wolfgang Huber http://www.dkfz.de/abt0840/whuber DKFZ Dep. Molecular Genome Analysis 69120 Heidelberg Germany

In the past (six months ago?), I noticed this type of problem too. What I found was that the read routines had gotten confused by differences in the encoding for newlines on PCs and Unix. In my case, the .cdf files were first transferred to a PC and then copied over to a Unix system. I first encountered this back when things were done by perl scripts, where the errant behavior was more explicit and could be easily debugged. However, I recently made the same mistake *again* with the C-based code, and the result was the symptoms you describe (Fool me once, shame on you; fool me twice, shame on me :-). Running it on a PC produced the expected results. In your case, was the .cdf file (or maybe the .cel file) read onto a PC and then transferred to Unix (or vice versa)? In the fine tradition of the R mailing list, I should now make a snide comment about you not giving us enough info, but I'll pass on that for now. cheers, dave Dr. David O. Nelson Lawrence Livermore National Laboratory Box 808, L-441 Livermore CA 94551 On Thu, 2002-06-20 at 04:27, Wolfgang Huber wrote: > > > read.celfile (from package affy version 1.1.0) keeps hanging with cel files exported from dChip version 1.2. I am not sure how standards-conform the dChip-export is, but in any case it would seem useful for read.celfile to produce an intelligible error message rather than just hanging. > > I managed to work around this by commenting the lines > ff@masks <- .Call("getIndexExtraFromCEL", ... > ff@outliers <- .Call("getIndexExtraFromCEL", ... > > > Wolfgang. > > > Dr. Wolfgang Huber > http://www.dkfz.de/abt0840/whuber > DKFZ > Dep. Molecular Genome Analysis > 69120 Heidelberg > Germany > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >

On Thu, Jun 20, 2002 at 12:47:13PM -0700, David O. Nelson wrote: > In the past (six months ago?), I noticed this type of problem too. What > I found was that the read routines had gotten confused by differences in > the encoding for newlines on PCs and Unix. In my case, the .cdf files > were first transferred to a PC and then copied over to a Unix system. I > first encountered this back when things were done by perl scripts, where > the errant behavior was more explicit and could be easily debugged. thanks for the pointer... but what puzzles me is that the Affymetrix imaging software generating the CEL from the images only exists (to my humble knowledge) for Windows... ...so technically all CEL files are coming from Windows... ...my wild guess would be that a local Perl hero messed up something in the files... (and as advertised before in this very same list the parser is designed to parse *Affymetrix* files rather than Affymetrix flavored files)... > > However, I recently made the same mistake *again* with the C-based code, > and the result was the symptoms you describe (Fool me once, shame on > you; fool me twice, shame on me :-). Running it on a PC produced the > expected results. > > In your case, was the .cdf file (or maybe the .cel file) read onto a PC > and then transferred to Unix (or vice versa)? > > In the fine tradition of the R mailing list, I should now make a snide > comment about you not giving us enough info, but I'll pass on that for > now. > ...too bad... I missed a chance to reply that contributions to improve things are also welcome... L. > cheers, > > dave > > Dr. David O. Nelson > Lawrence Livermore National Laboratory > Box 808, L-441 > Livermore CA 94551 > > > On Thu, 2002-06-20 at 04:27, Wolfgang Huber wrote: > > > > > > read.celfile (from package affy version 1.1.0) keeps hanging with cel files exported from dChip version 1.2. I am not sure how standards-conform the dChip-export is, but in any case it would seem useful for read.celfile to produce an intelligible error message rather than just hanging. > > > > I managed to work around this by commenting the lines > > ff@masks <- .Call("getIndexExtraFromCEL", ... > > ff@outliers <- .Call("getIndexExtraFromCEL", ... > > > > > > Wolfgang. > > > > > > Dr. Wolfgang Huber > > http://www.dkfz.de/abt0840/whuber > > DKFZ > > Dep. Molecular Genome Analysis > > 69120 Heidelberg > > Germany > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor@stat.math.ethz.ch > > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor > > > > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor

Hi David, thanks for your suggestion. As Laurent suggested, it's not the DOS/Unix thing, but something else: Indeed, the CEL files I was having trouble with were written from dChip on a Windos machine, while I read them into R on a Unix machine. And indeed, the CEL files do contain DOS-type newlines. However, that is not the problem: - I used other CEL files (e.g., http://www- genome.wi.mit.edu/mpr/lung/) that also have DOS-newlines, and with whom read.celfile works fine. - I did remove the extra ^M (LF) character, both with emacs and with dos2unix; that did not help. Taking a closer look at the problematic CEL files, they have no sections headed [MASKS], [OUTLIERS]; while "normal" CEL files seem to have these sections. So may I guess that the problem lies in here - read.celfile expects these keywords in the CEL files, and when it doesn't find them, it hangs. I guess that could be easily fixed. > In the fine tradition of the R mailing list, I should now make a snide > comment about you not giving us enough info, but I'll pass on that for > now. Right. I wasn't sure whether this would be an acceptable message to this mailing list at all, since the troubling CEL files weren't written by Affy software and one cannot expect the parser to cope with all sorts of other stuff. So, my message came out in a sort of half-hearted manner... Anyway, being rather unsophisticated with Affy and CEL files, I just didn't know what more to supply, short of attaching the whole CEL file :) Best regards Wolfgang ----------------------------------- Dr. Wolfgang Huber http://www.dkfz.de/abt0840/whuber DKFZ Dep. Molecular Genome Analysis 69120 Heidelberg Germany