[phenixbb] phenix.refine crash -- other error

Ulrich Baumann ulrich.baumann at ibc.unibe.ch
Wed Mar 14 07:49:11 PDT 2007


Dear Peter,
many thanks! Will try this -- I do not know about the FFT algorithm in  
phenix, but in CNS you could specify memory use and do calculations on a  
subgrid. Thi stakes more time but needs less memory. There is no cuh thing  
in phenix?


many thanks again for your prompt help,

Ulrich

On Tue, 13 Mar 2007 20:07:46 +0100, Peter Zwart <PHZwart at lbl.gov> wrote:

> Hi Ulrich,
>
> You could try refining your structure in a primitive setting, this might
>   alleviate memory usage a bit (not sure though).
>
> try
>
> phenix.python $MMTBX_DIST/mmtbx/examples/reindex.py data.file=mydata.mtz
>   model.file=mymodel.pdb action=reindex standard_laws=niggli
>
> You will get a new mtz file and a new pdb file. Unfortunately, the free
> flags are not copied over, so you would have to reindex those separately
> using iotbx.reflection_file_converter and read that file in separately
> in phenix.refine.
>
> This might be a bit criptic, if so, let me know and I will post more
> details.
>
> I am not sure this does the trick though, I hope Ralf or Pavel can give
> some suggestions.
>
> Cheers
>
> Peter
>
>
>
>
> Ulrich Baumann wrote:
>> Hi there,
>>
>> if you remove the ANISOU card from the input pdb file, phenix will run  
>> fine
>> ....sometimes not.
>>
>> phenix.refine ( Version: 2007_01_24_2154) crashed for me as well: we  
>> have a
>> rather large unit cell and good resolution. The program crashes  
>> reproducibly
>> (see below) when using resolution limits better than 3 A, although  
>> sometimes
>> this only occurs in the 2nd macrocycle.
>>
>> Working crystal symmetry after inspecting all inputs:
>>   Unit cell: (202.401, 202.401, 317.731, 90, 90, 120)
>>   Space group: R 3 :H (No. 146)
>>
>>
>> We have 2 GB memory and could stuff in more but maybe we can limit the  
>> memeory
>> use in a different way?
>>
>>
>> Thanks a lot for your help,
>>
>> Ulrich
>>
>>
>> ============================= ml refinement start  
>> =============================
>>
>>
>>    ----------structure factors based statistics (before  
>> refinement)----------
>>
>> Traceback (most recent call last):
>>   File  
>> "/usr/local/cci_apps/cci_apps_sources/phenix/phenix/command_line/refine.p
>> y", line 5, in <module>
>>     command_line.run(command_name="phenix.refine", args=sys.argv[1:])
>>   File  
>> "/usr/local/cci_apps/cci_apps_sources/phenix/phenix/refinement/command_li
>> ne.py", line 75, in run
>>     call_back_after_monitor_collect=call_back_after_monitor_collect)
>>   File  
>> "/usr/local/cci_apps/cci_apps_sources/phenix/phenix/refinement/driver.py"
>> , line 1108, in run
>>     log                     = log)
>>   File  
>> "/usr/local/cci_apps/cci_apps_sources/phenix/phenix/refinement/strategies
>> .py", line 174, in refinement_machine
>>     abcd              = abcd)
>>   File "/usr/local/cci_apps/cci_apps_sources/mmtbx/mmtbx/f_model.py",  
>> line 185,
>> in __init__
>>     b_cart               = b_cart)
>>   File "/usr/local/cci_apps/cci_apps_sources/mmtbx/mmtbx/f_model.py",  
>> line 466,
>> in update_xray_structure
>>     f_mask = bulk_solvent_mask_obj.structure_factors(miller_set=  
>> self.f_obs)
>>   File "/usr/local/cci_apps/cci_apps_sources/mmtbx/mmtbx/masks.py",  
>> line 97, in
>> structure_factors
>>     flex.grid(fft_manager.m_real()).set_focus(fft_manager.n_real()))
>> MemoryError
>>
>> ----------------------------------------------------------
>> -------------------------------------------------------
>> Quoting Pavel Afonine <pafonine at lbl.gov>:
>>
>>> Hi,
>>>
>>> this is known problem and it will be fixed ASAP; sorry for this.
>>> This presumably happens because an atom selected to be refined as
>>> individual anisotropic is included into a TLS group.
>>> I will let you know once the fixed version of CCI Apps is available.
>>>
>>> Pavel.
>>>
>>>
>>>
>>> Jianghai Zhu wrote:
>>>> Hi,
>>>>
>>>> I was using phenix.refine from the latest cci apps bundle to do
>>>> tls+individual_sites+individual_adp refinement.  The program crashed
>>>> with the following error message.
>>>>
>>>>              ----------Target weights (before refinement)----------
>>>>
>>>>
>>>> Traceback (most recent call last):
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/phenix/phenix/command_line/refine.py",
>>>> line 5, in <module>
>>>>     command_line.run(command_name="phenix.refine", args=sys.argv[1:])
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/phenix/phenix/refinement/command_line.py",
>>>> line 75, in run
>>>>     call_back_after_monitor_collect=call_back_after_monitor_collect)
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/phenix/phenix/refinement/driver.py",
>>>> line 1108, in run
>>>>     log                     = log)
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/phenix/phenix/refinement/strategies.py",
>>>> line 245, in refinement_machine
>>>>     log                           = log)
>>>>   File
>>>>
>>> "/nfs/home/jzhu/cci_apps_sources/phenix/phenix/refinement/weight_xray_chem.py",
>>>
>>>> line 222, in __init__
>>>>     compute_gradients = True)
>>>>   File "/nfs/home/jzhu/cci_apps_sources/mmtbx/mmtbx/restraints.py",
>>>> line 144, in energies_adp_aniso
>>>>     xray_structure = xray_structure,
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/cctbx/cctbx/adp_restraints/__init__.py",
>>>> line 109, in __init__
>>>>     self.check_flags(fl_i)
>>>>   File
>>>> "/nfs/home/jzhu/cci_apps_sources/cctbx/cctbx/adp_restraints/__init__.py",
>>>> line 177, in check_flags
>>>>     assert not fl.use_u_aniso()
>>>> AssertionError
>>>>
>>>> Anyone has any idea what is happening?  Thanks.
>>>>
>>>> Jianghai
>>>>
>>>> +++++++++++++++++++++++++++++++
>>>> Jianghai Zhu, Ph.D
>>>> CBR Institute for Biomedical Research
>>>> Department of Pathology
>>>> Harvard Medical School
>>>> 200 Longwood Ave., Boston, MA 02115
>>>> Ph: 617-278-3211
>>>> Fx: 618-278-3232
>>>> +++++++++++++++++++++++++++++++
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> phenixbb mailing list
>>>> phenixbb at phenix-online.org
>>>> http://www.phenix-online.org/mailman/listinfo/phenixbb
>>>>
>>
>>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb



-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/




More information about the phenixbb mailing list