Convertion fail

[smoretti]

Hi

I am new to sbfc and try to convert an SBML model to Octave or Matlab for example.
I tried with my own models as well as with examples/SBML.xml but I always got the same error message in log files:

./sbfConverter.sh SBMLModel SBML2Octave examples/SBML.xml

2017-01-13 13:22
sbfConverter.sh: Convertion, using SBML2Octave, for 'examples/SBML.xml'...
------------------------------------------------------------
miriam.xml.export=/home/.../miriam.xml: DATALIMIT value should be a positive integer. Job not submitted.

Could help me to figure out what happens?

Regards

[smoretti]

Sorry
I use version 1.3.7, downloaded from sourceforge (zip file)

miriam.xml is there. I have updated it with the updateMiriamRegistry.sh script but it fails as before the update.

[Nicolas Rodriguez]

Which version are you using ? Is it an archive that you downloaded from sourceforge or are you using an svn checkout and the sources ?

One thing to check is that you have a file named miriam.xml and that it is proper XML.

Thanks,
Nico

[Nicolas Rodriguez]

It does seems to be working fine for me when using the zip file.

Which version of java are you using ? What does the command 'java -version' returns ?

And 'lsb_release -a' ?

Thanks,
Nico

[smoretti]

I use Java 1.8.0_112 from Sun/Oracle

java version "1.8.0_112"
Java(TM) SE Runtime Environment (build 1.8.0_112-b15)
Java HotSpot(TM) 64-Bit Server VM (build 25.112-b15, mixed mode)

'lsb_release -a' returns:

LSB Version:    :base-4.0-amd64:base-4.0-noarch:core-4.0-amd64:core-4.0-noarch:graphics-4.0-amd64:graphics-4.0-noarch:printing-4.0-amd64:printing-4.0-noarch
Distributor ID:    CentOS
Description:    CentOS release 6.8 (Final)
Release:    6.8
Codename:    Final

[Nicolas Rodriguez]

Strange, that look fine.

Can you edit the 'sbfConverterOnline.sh' script and comments the lines messing with the PATH and LIBRARY_PATH, it should look like that:

---------------------
# Setting up libsbml, needed for the SBML2SBML converters
#LD_LIBRARY_PATH=/nfs/production/biomodels/sw/libSBML-5.11.6-Linux/usr/lib64

# Setting up java
#export JAVA_HOME=/nfs/public/rw/webadmin/java/jdks/latest_1.8
#export PATH=${JAVA_HOME}/bin:${PATH}

---------------------

Then add this line before the last 'eval' line:

    echo  eval $COMMAND -Dmiriam.xml.export=${SBF_CONVERTER_HOME}/miriam.xml org.sbfc.converter.Converter $MODEL_NAME $CONVERTER_NAME $SBML_DIR >> $LOG_FILE  2>&1

Then try again the conversion, using:

./sbfConverterOnline.sh SBMLModel SBML2Octave examples/SBML.xml

And send me the full content of the log file.

Thanks,
Nico

--
You received this message because you are subscribed to the Google Groups "sbfc-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbfc-forum+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/sbfc-forum/03d2c5b8-f98d-46eb-988c-aafb2b4e110c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[smoretti]

Looks to work now, I have a examples/SBML.m file.

Here is the log:

$ cat log/SBML.xml/SBML.xml-SBML2Octave-export-2017-01-13.log
------------------------------------------------------------
2017-01-13 15:22
sbfConverterOnline.sh: Convertion, using SBML2Octave, for 'examples/SBML.xml'...
------------------------------------------------------------
eval java -Dmiriam.xml.export=/home/.../miriam.xml org.sbfc.converter.Converter SBMLModel SBML2Octave examples/SBML.xml
Loading the Identifiers.org registry database from  '/home/.../miriam.xml'.
SBML2Octave : Export in progress...
Reaction : id = veq
compartment_cytosol*(reaction_veq_kon*x(4)*x(3)-reaction_veq_koff*x(1))
Reactant flux map : id = E
Reactant flux map : id = S
Reaction : id = vcat
compartment_cytosol*reaction_vcat_kcat*x(1)
Reactant flux map : id = ES
false
Species : id = ES
 (1/(compartment_cytosol))*(( 1.0 * reaction_veq) + (-1.0 * reaction_vcat))
false
Species : id = P
 (1/(compartment_cytosol))*(( 1.0 * reaction_vcat))
false
Species : id = S
 (1/(compartment_cytosol))*((-1.0 * reaction_veq))
false
Species : id = E
 (1/(compartment_cytosol))*((-1.0 * reaction_veq) + ( 1.0 * reaction_vcat))

[Nicolas Rodriguez]

On 13/01/17 14:25, smoretti wrote:

Looks to work now, I have a examples/SBML.m file.

Might be the modification to the PATH or LD_LIBRARY_PATH but it should not matter.

Do you have the 'bsub' command available on the shell ? If you do, the first script is trying to make use of it
to send the job to run on a cluster.

--
You received this message because you are subscribed to the Google Groups "sbfc-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbfc-forum+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/sbfc-forum/9eb0019d-e15f-4bd7-8f0a-15e21cb16df2%40googlegroups.com.

[smoretti]

I have bsub.

Should I comment all lines with PATH or LD_LIBRARY_PATH in all scripts to make it work for me?

[Nicolas Rodriguez]

On 13/01/17 15:58, smoretti wrote:

I have bsub.

I think the main problem is bsub (although, you can also comment the PATH or LD_LIBRARY_PATH lines). Is it a LSF related
command ?

The problematic lines would be:

if [ "`which bsub 2> /dev/null`" == "" ] ; then
    COMMAND="java $PROPERTIES"
fi

if you comment the 'if' and 'fi' lines (as they are in the 'online' script), then it will always use java instead of trying to launch the job on the cluster using 'bsub'.

We should probably clean those scripts before release as they are customized to run on our cluster at EBI or at least check that 'bsub -V' return a string that contain LSF in the script if bsub is found.

cheers,
Nico

--
You received this message because you are subscribed to the Google Groups "sbfc-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbfc-forum+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/sbfc-forum/b501af39-0d3e-4745-8ac2-7ef300b49142%40googlegroups.com.

[smoretti]

I have bsub.

I think the main problem is bsub (although, you can also comment the PATH or LD_LIBRARY_PATH lines). Is it a LSF related
command ?

This is a LSF related command but it is not fully functional on desktop computers compared to our cluster nodes.

So I replaced all

COMMAND="bsub .*? java

by

COMMAND="java

This way sbfc should work on all our computers. And people can still submit with bsub externally.

Regards
Thanks
 

[Nicolas Rodriguez]

On 16/01/17 09:42, smoretti wrote:

I have bsub.

I think the main problem is bsub (although, you can also comment the PATH or LD_LIBRARY_PATH lines). Is it a LSF related
command ?

This is a LSF related command but it is not fully functional on desktop computers compared to our cluster nodes.

So I replaced all

COMMAND="bsub .*? java

by

COMMAND="java

This way sbfc should work on all our computers. And people can still submit with bsub externally.

Yes, exactly.

 

The problematic lines would be:

if [ "`which bsub 2> /dev/null`" == "" ] ; then
    COMMAND="java $PROPERTIES"
fi

if you comment the 'if' and 'fi' lines (as they are in the 'online' script), then it will always use java instead of trying to launch the job on the cluster using 'bsub'.

We should probably clean those scripts before release as they are customized to run on our cluster at EBI or at least check that 'bsub -V' return a string that contain LSF in the script if bsub is found.

cheers,
Nico

Should I comment all lines with PATH or LD_LIBRARY_PATH in all scripts to make it work for me?

Looks to work now, I have a examples/SBML.m file.

Might be the modification to the PATH or LD_LIBRARY_PATH but it should not matter.

Do you have the 'bsub' command available on the shell ? If you do, the first script is trying to make use of it
to send the job to run on a cluster.

--
You received this message because you are subscribed to the Google Groups "sbfc-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbfc-forum+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/sbfc-forum/15ac04d1-d473-408f-ad21-c68dd1e8e426%40googlegroups.com.

[smoretti]

I got another issue with log directory.
When installed as root, the other users cannot write in the log directory.
I fixed that by changing all occurrences of

LOG_FILE_FOLDER=${SBF_CONVERTER_HOME}/log/...

by

LOG_FILE_FOLDER=.

The same thing happens when the file to convert is in a read-only folder, the converted file cannot be written.
Is there a way to change this behaviour (without copying the file to convert in a read-write place first)?

[Nicolas Rodriguez]

We haven't implemented any options to do so at the moment but in fact, it would be easy to do
as we already have a method that take the output file name that could be used:

org.sbfc.converter.Converter.convertFromFile(String, String, String, String)

But it would require to modify the code of the Converter class, although if you use the GUI (with sbfConverterGUI.sh)
you can already specify the output file name.

For the scripts, we could add a fourth optional arguments that would be the output file name/path, would that be fine ?

Thanks,
Nico

[smoretti]

The same thing happens when the file to convert is in a read-only folder, the converted file cannot be written.
Is there a way to change this behaviour (without copying the file to convert in a read-write place first)?

We haven't implemented any options to do so at the moment but in fact, it would be easy to do
as we already have a method that take the output file name that could be used:

org.sbfc.converter.Converter.convertFromFile(String, String, String, String)

But it would require to modify the code of the Converter class, although if you use the GUI (with sbfConverterGUI.sh)
you can already specify the output file name.

For the scripts, we could add a fourth optional arguments that would be the output file name/path, would that be fine ?

Yes, that would be find this way.
Thanks
 

Thanks,
Nico