GLU crash

[INK]

Hello,
we would much appreciate any help on the error we are getting:

Traceback: Traceback (most recent call last):
File "/home/apps/GLU/2012-03-15/lib/python2.7/site-packages/
glu-1.0b3_dev-py2.7-linux-x86_64.egg/glu/lib/glu_launcher.py", line
221, in main
progmain()
File "/home/apps/GLU/2012-03-15/lib/python2.7/site-packages/
glu-1.0b3_dev-py2.7-linux-x86_64.egg/glu/modules/assoc/logit1.py",
line 195, in main
null.fit()
File "/home/apps/GLU/2012-03-15/lib/python2.7/site-packages/
glu-1.0b3_dev-py2.7-linux-x86_64.egg/glu/lib/glm/glm.py", line 1236,
in fit
beta,ss,W,resids,rank,s,vt =
linear_least_squares(XX,zz,cond=COND)
File "/home/apps/GLU/2012-03-15/lib/python2.7/site-packages/
glu-1.0b3_dev-py2.7-linux-x86_64.egg/glu/lib/glm/glm.py", line 296, in
linear_least_squares
overwrite_a=0, overwrite_b=0)
ValueError: too many values to unpack

This is running "glu assoc.logit1" using a fresh build, not really
well tested (glu's own test crashed)
OS: SLES11 SP1
version: 1.0b3 fetched from repo on 2012-03-15.
All builds are using GNU 4.3.4

req installed
HDF5 1.8.8
ACML 4.4.0
SQLite 3.5.9 3.6.4

Python 2.7.2 -
setuptools 0.6c11 -
NumPy 1.1.0 1.6.1
SciPy 0.9.0 0.10.1
numexpr 1.4.2 1.4.2
PyTables 2.0.4 2.3.1
h5py 2.0 2.0.1
Ply 2.5 3.4
pysam 0.5 0.6
BioPython 1.58 1.59
matplotlib 1.0.1 1.1.0
Cython 0.15 0.15.1
xlrd 0.7.1 -
xlwt 0.7.2 0.7.3
cvxopt 1.1.1 1.1.4

thanks
igor

[Jacobs, Kevin (NIH/NCI) [C]]

I have not tested GLU with that mix of version of NumPy, SciPy and ACML. I'll do some poking around and will get back to you shortly.

Best regards,
-Kevin

thanks
igor

--
You received this message because you are subscribed to the Google Groups "glu-users" group.
To post to this group, send email to glu-...@googlegroups.com.
To unsubscribe from this group, send email to glu-users+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/glu-users?hl=en.

[INK]

Kevin,
many thanks for looking into this.
How likely is the error due to NumPy?
What versions of NumPy, SciPy and ACML would you suggest to downgrade
to?
thank you
Igor

On Mar 20, 12:19 pm, "Jacobs, Kevin (NIH/NCI) [C]"

[Kevin Jacobs <jacobs@bioinformed.com>]

I'm still using 0.9.0, but I don't think you should downgrade.  I'll have a fix checked in this afternoon that will make the code forward and backward compatible.  Would you mind testing it for me once it is ready, since it is complicated for me to update my local installation at the moment?

-Kevin

[INK]

Yes, absolutely. I'm happy to run whatever is necessary to the ensure
the code is working correctly.
Are there other tests than the one which comes with the source?

On Mar 21, 4:29 pm, "Kevin Jacobs <jac...@bioinformed.com>"

[Jacobs, Kevin (NIH/NCI) [C]]

I have an internal test suite, but it uses datasets that I cannot share publically. It is certainly a major goal of mine to have a better test suite that can also serve as examples of how best to use GLU.

I've just checked in code that should fix the problem that you observed. Can you get a fresh source tree using the Mercurial repository on the Google Code site and give it a try?

Thanks,

[INK]

Kevin, many thanks!
I was on r1071 and updating to r1072 did the trick. The job that
previously crashed run successfully and the output seems correct.

I should say however that GLU's own test still crashed. Perhaps the
output below can be of some help.

python2.7 setup.py test
...
< cut >
...
/usr/bin/gfortran -Wall -Wall -shared build/temp.linux-x86_64-2.7/
build/src.linux-x86_64-2.7/glu/lib/glm/_glmnetmodule.o build/
temp.linux-x86_64-2.7/build/src.linux-x86_64-2.7/fortranobject.o build/
temp.linux-x86_64-2.7/glu/lib/glm/GLMnet.o -L/home/apps/python/2.7.2/
lib -lpython2.7 -lgfortran -o glu/lib/glm/_glmnet.so
running scons
/home/apps/python/2.7.2/lib/python2.7/site-packages/nose/config.py:
259: RuntimeWarning: Option 'with-doctest' in config file 'setup.cfg'
ignored: excluded by runtime environment
warn(msg, RuntimeWarning)

----------------------------------------------------------------------
Ran 0 tests in 0.006s

OK
Error in atexit._run_exitfuncs:

Traceback (most recent call last):

File "/home/apps/python/2.7.2/lib/python2.7/atexit.py", line 24, in
_run_exitfuncs
func(*targs, **kargs)
File "/home/apps/python/2.7.2/lib/python2.7/multiprocessing/
util.py", line 284, in _exit_function
info('process shutting down')
TypeError: 'NoneType' object is not callable
Error in sys.exitfunc:

Traceback (most recent call last):

File "/home/apps/python/2.7.2/lib/python2.7/atexit.py", line 24, in
_run_exitfuncs
func(*targs, **kargs)
File "/home/apps/python/2.7.2/lib/python2.7/multiprocessing/
util.py", line 284, in _exit_function
info('process shutting down')
TypeError: 'NoneType' object is not callable


Many thanks again for your help!

Igor



On Mar 21, 5:49 pm, "Jacobs, Kevin (NIH/NCI) [C]"

[Kevin Jacobs <jacobs@bioinformed.com>]

Great news!  Can you try to run the test via:

  python2.7 setup.py nosetests

That is how I do it locally.  Some of the tests are expected to fail, mainly due to differences in math libraries and because I use the rpy interface to R to validate some of my code.

-Kevin

[INK]

On Mar 22, 2:56 pm, "Kevin Jacobs <jac...@bioinformed.com>"

<bioinfor...@gmail.com> wrote:
> Great news!  Can you try to run the test via:
>
>   python2.7 setup.py nosetests
>
> That is how I do it locally.  Some of the tests are expected to fail,
> mainly due to differences in math libraries and because I use the rpy
> interface to R to validate some of my code.
>
> -Kevin

Done.

Ran 299 tests in 1.653s

FAILED (errors=2, failures=11)

Do you want to see the whole output?

[Jacobs, Kevin (NIH/NCI) [C]]

Nope-- that is what I expected to see.

Thanks for your help. Please let me know if you have any questions or have any more problems.

-Kevin

-----Original Message-----
From: INK [mailto:i.n....@googlemail.com]
Sent: Thursday, March 22, 2012 12:01 PM
To: glu-users
Subject: [glu-users] Re: GLU crash

Done.

FAILED (errors=2, failures=11)

[INK]

If anyone, I should be thanking you for the code and swift support!

Igor

On Mar 22, 4:12 pm, "Jacobs, Kevin (NIH/NCI) [C]"