On Wed, Apr 25, 2012 at 11:48 AM, Dan Tenenbaum <dtenenba@fhcrc.org> wrote:
On Wed, Apr 25, 2012 at 9:53 AM, Dan Tenenbaum <dtenenba@fhcrc.org> wrote:
Hello Hans-Rudolph,
On Wed, Apr 25, 2012 at 1:37 AM, Hans-Rudolf Hotz <hrh@fmi.ch> wrote:
Hi Dan
There seems to be several issues connected with each other or not, I don't know.....
- Let's start with the 'curiosity': Do you get this problem with any tool?
No. And, I should be a bit more specific about what happens. When I click "execute" I see this:
The following job has been successfully added to the queue:
36: RNASeq tool on data 3 and data 4 37: RNASeq tool on data 3 and data 4
Does it also happen with a 'simple' (ie not using R) tool you
add?
No.
- When you execute your R script on the command line, are you running it as the same user as Galaxy executes the job?
Yes.
- we execute R scripts the following way:
<command>Rscript --vanilla /full path to script/script.R -n $name -i $infile -o $outfile</command>
and in the R script we use the library 'getopt'
Good idea. Taking a closer, look, I found my R script was not expecting arguments in the same order that the XML script provided them in. I fixed that and now the R script runs without warning or error, but I still get the "8-bit bytestrings" error, and the error "Unable to finish job".
However, by printing out the arguments within R, I can see the name of the output files that are supposed to be generated, and it looks like they are generated. So why doesn't the job finish?
Sorry for the noise of answering my own question, which I find has been asked and answered before, but perhaps this will help someone. Here's the answer: http://lists.bx.psu.edu/pipermail/galaxy-dev/2010-August/003161.html Basically, if any output goes to STDERR, galaxy considers that the job failed. The link above suggests to do this: jobcommand ARGS 2>/dev/null But I changed that to: jobcommand ARGS 2>&1 that way, STDERR is redirected to STDOUT and any important diagnostic information originally printed to STDERR is not lost. I added this to my Rscript_wrapper.sh script and all is well. Thanks, Dan
I think the 8 bit bytestrings error might be because R outputs backticks (`) and they show up escaped in the galaxy error like this: \xe2\x80\x98.
Is there some way that I can set my LOCALE so that R uses an encoding that galaxy is happier with? Or perhaps I can get R to stop outputting those backticks somehow.
Answering my own question here (but following up with more questions). R was ouputting not backticks but curly quotes, and I turned these off by putting:
options("useFancyQuotes" = FALSE)
in my R script.
However, my R script is still failing:
Dataset generation errors
Dataset 46: RNASeq tool on data 3 and data 4
Tool execution generated the following error message: BiocInstaller version 1.5.7, ?biocLite for help Loading required package: org.Dm.eg.db Loading required package: AnnotationDbi Loading required package: BiocGenerics
Attaching package: 'BiocGenerics'
The following object(s) are masked from 'package:stats':
xtabs
The following object(s) are masked from 'package:base':
anyDuplicated, cbind, colnames, duplicated, eval, Filter, Find, get, intersect, lapply, Map, mapply, mget, order, paste, pmax, pmax.int, pmin, pmin.int, Position, rbind, Reduce, rep.int, rownames, sapply, setdiff, table, tapply, union, unique
Loading required package: Biobase Welcome to Bioconductor
Vignettes contain introductory material; view with 'browseVignettes()'. To cite Bioconductor, see 'citation("Biobase")', and for packages 'citation("pkgname")'.
Loading required package: DBI
Calculating library sizes from column totals. Running estimateCommonDisp() on DGEList object before proceeding with estimateTagwiseDisp(). The tool produced the following additional output: pdf_outfile= /Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_46.dat csv_outfile= /Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_47.dat pdf 2
The thing is, none of this is an error. This is the exact output the script is supposed to produce, and it prints out the filenames of the two output files it creates, and I can verify that those exist and have the correct contents. So why does it say there is an error?
Could it be the return code? I added q(save="no", status=0) to the end of the R script to ensure that the script exits with status code 0, but that did not change the error I got.
So I am not sure how to solve this. The script works exactly as it's supposed to and creates the desired output, but tells me there is an error, and gives no way through the web interface to retrive the created output file.
Also, I found out why two jobs were started each time I click "Execute". It's because I had two files in my <outputs> section:
<outputs> <data format="pdf" name="out_file1" /> <data format="csv" name="out_file2" /> </outputs>
Because my script generates two output files, a csv and a PDF. I took out the first <data> line above, and stopped passing the corresponding argument to my script, and the script no longer creates the PDF. Now the job only runs once each time I click "Execute". But I would like the script to once again produce both output files.
So, perhaps it was doing the right thing all along...showing one dataset in the history panel for each output file produced.
Looking through the galaxy distribution, I see at least one other tool that does the same thing. So, sorry for the noise on this.
In any case, I still get the same error described above even when only one output file is produced.
Thanks, Dan
I hope this get's you a little bit further.
Thanks much, Dan
Regards, Hans
On 04/24/2012 11:56 PM, Dan Tenenbaum wrote:
Apologies for originally posting this to galaxy-user; now I realize it belongs here.
Hello,
I'm a galaxy newbie and running into several issues trying to adapt an R script to be a galaxy tool.
I'm looking at the XY plotting tool for guidance (tools/plot/xy_plot.xml), but I decided not to embed my script in XML, but instead have it in a separate script file, that way I can still run it from the command line and make sure it works as I make incremental changes. (So my script starts with args<- commandArgs(TRUE)). Also, if it doesn't work, this suggests to me that there is a problem with my galaxy configuration.
First, I tried using the r_wrapper.sh script that comes with the XY plotting tool, but it threw away my arguments:
An error occurred running this job: ARGUMENT '/Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_4.dat' __ignored__
ARGUMENT '/Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_3.dat' __ignored__
ARGUMENT 'Fly' __ignored__
ARGUMENT 'Tagwise' __ignored__
etc.
So then I tried just switching to Rscript:
<command interpreter="bash">Rscript RNASeq.R $countsTsv $designTsv "$organism" $dispersion $minimumCountsPerMillion $minimumSamplesPerTranscript $out_file1 $out_file2</command>
(My script produces as output a csv file and a pdf file. The final two arguments I'm passing are the names of those files.)
But then I get an error that Rscript can't be found.
So I wrote a little wrapper script, Rscript_wrapper.sh:
#!/bin/sh
Rscript $*
And called that: <command interpreter="bash">Rscript_wrapper.sh RNASeq.R $countsTsv $designTsv "$organism" $dispersion $minimumCountsPerMillion $minimumSamplesPerTranscript $out_file1 $out_file2</command>
Then I got an error that RNASeq.R could not be found.
So then I added the absolute path to my R script to the<command> tag. This seemed to work (that is, it got me further, to the next error), but I'm not sure why I had to do this; in all the other tools I'm looking at, the directory to the script to run does not have to be specified; I assumed that the command would run in the appropriate directory.
So now I've specified the full path to my R script:
<command interpreter="bash">Rscript_wrapper.sh /Users/dtenenba/dev/galaxy-dist/tools/bioc/RNASeq.R $countsTsv $designTsv "$organism" $dispersion $minimumCountsPerMillion $minimumSamplesPerTranscript $out_file1 $out_file2</command>
And I get the following long error, which includes all of the output of my R script:
Traceback (most recent call last): File "/Users/dtenenba/dev/galaxy-dist/lib/galaxy/jobs/runners/local.py", line 133, in run_job job_wrapper.finish( stdout, stderr ) File "/Users/dtenenba/dev/galaxy-dist/lib/galaxy/jobs/__init__.py", line 725, in finish self.sa_session.flush() File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/scoping.py", line 127, in do return getattr(self.registry(), name)(*args, **kwargs) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/session.py", line 1356, in flush self._flush(objects) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/session.py", line 1434, in _flush flush_context.execute() File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/unitofwork.py", line 261, in execute UOWExecutor().execute(self, tasks) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/unitofwork.py", line 753, in execute self.execute_save_steps(trans, task) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/unitofwork.py", line 768, in execute_save_steps self.save_objects(trans, task) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/unitofwork.py", line 759, in save_objects task.mapper._save_obj(task.polymorphic_tosave_objects, trans) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/orm/mapper.py", line 1413, in _save_obj c = connection.execute(statement.values(value_params), params) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/engine/base.py", line 824, in execute return Connection.executors[c](self, object, multiparams, params) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/engine/base.py", line 874, in _execute_clauseelement return self.__execute_context(context) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/engine/base.py", line 896, in __execute_context self._cursor_execute(context.cursor, context.statement, context.parameters[0], context=context) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/engine/base.py", line 950, in _cursor_execute self._handle_dbapi_exception(e, statement, parameters, cursor, context) File "/Users/dtenenba/dev/galaxy-dist/eggs/SQLAlchemy-0.5.6_dev_r6498-py2.7.egg/sqlalchemy/engine/base.py", line 931, in _handle_dbapi_exception raise exc.DBAPIError.instance(statement, parameters, e, connection_invalidated=is_disconnect) ProgrammingError: (ProgrammingError) You must not use 8-bit bytestrings unless you use a text_factory that can interpret 8-bit bytestrings (like text_factory = str). It is highly recommended that you instead just switch your application to Unicode strings. u'UPDATE job SET update_time=?, stdout=?, stderr=? WHERE job.id = ?' ['2012-04-24 18:55:45.791417', '', 'BiocInstaller version 1.5.7, ?biocLite for help\nWarning message:\nNAs introduced by coercion \nLoading required package: methods\nLoading required package: limma\nLoading required package: BiasedUrn\nLoading required package: geneLenDataBase\nLoading required package: org.Dm.eg.db\nLoading required package: AnnotationDbi\nLoading required package: BiocGenerics\n\nAttaching package: \xe2\x80\x98BiocGenerics\xe2\x80\x99\n\nThe following object(s) are masked from \xe2\x80\x98package:stats\xe2\x80\x99:\n\n xtabs\n\nThe following object(s) are masked from \xe2\x80\x98package:base\xe2\x80\x99:\n\n anyDuplicated, cbind, colnames, duplicated, eval, Filter, Find,\n get, intersect, lapply, Map, mapply, mget, order, paste, pmax,\n pmax.int, pmin, pmin.int, Position, rbind, Reduce, rep.int,\n rownames, sapply, setdiff, table, tapply, union, unique\n\nLoading required package: Biobase\nWelcome to Bioconductor\n\n Vignettes contain introductory material; view with\n \'browseVignettes()\'. To cite Bioconductor, see\n \'citation("Biobase")\', and for packages \'citation("pkgname")\'.\n\nLoading required package: DBI\n\nCalculating library sizes from column totals.\nError in matrix(u, nrow = nrows, byrow = TRUE) : \n negative extents to matrix\nCalls: plotMDS.DGEList ... equalizeLibSizes -> splitIntoGroups -> lapply -> FUN -> matrix\nExecution halted\n', 15]
Note that if I run my script from the command line:
./Rscript_wrapper.sh RNASeq.R /Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_4.dat /Users/dtenenba/dev/galaxy-dist/database/files/000/dataset_3.dat Fly 1 1 Tagwise MDSPlot.pdf outputs.csv
It works fine and does not produce a warning about "NAs introduced by coercion", nor does it fail with the "Error in matrix" above.
So, can anyone tell me what is going wrong here? Why does R behave differently in galaxy than it does on the command line? (I'm using the same instance of R, same machine, for my galaxy and command-line efforts). Is this 8-bit bytestring error a red herring? Can I filter it so that galaxy is happy?
Finally, one other curiosity. Every time I hit "Execute" in galaxy to run my tool, it is run twice--two jobs are created (which each fail in the same way). Why is this?
My R script: https://gist.github.com/2482783
My XML file: https://gist.github.com/2482792
I can share more data (such as sample input files) if necessary.
Thanks for your help. Dan ___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: