Hello,

It works, Relion runs.
Thanks for your help


Le 29/05/13 19:09, Sjors Scheres a écrit :
> Hi Aurelie,
>
> The recommended way of preprocessing is through the preprocessing 
> run_type in the GUI.... For example, how was "previous ctf correction" 
> done? Hopefully nothing more than a standard phase flipping? Anyway, 
> re-running the preprocessing in RELION is probably very quick AND 
> guarantees that things are done in the optimal way for relion.
>
> That said, the mistake lies elsewhere in this case: your particles are 
> 128x128, yet you use a RADIUS for the background normalization of 90 
> pixels.... You probably mean 45.
>
> Hope that helps,
> S
>
>
>
> On 05/29/2013 05:53 PM, Aurélie Nicolas wrote:
>> I think the problem come from the normalisation.
>> I tried to open my normalized particles with xmipp and all my images 
>> are black.
>>
>> My command line is : mpirun -np 24 -machinefile hosts.mpi 
>> /opt/relion-1.1/bin/relion_preprocess_mpi --operate_on 
>> all_images_ctf.sta
>> r --o test_norm --norm --bg_radius 90
>>
>> I have 53 mrcs stacks (1 stack / image) with previously ctf 
>> corrected. The size of particles are 128*128 px
>>
>> The beginning of my all_images_ctf.star file is :
>>
>> data_
>> loop_
>> _rlnImageName
>> _rlnMicrographName
>> _rlnDefocusU
>> _rlnVoltage
>> _rlnSphericalAberration
>> _rlnAmplitudeContrast
>> [log in to unmask] 01 19999.54  200 2 0.07
>> [log in to unmask] 01 19999.54  200 2 0.07
>> ...
>> [log in to unmask] 08 25562.33  200 2 0.07
>> [log in to unmask] 08 25562.33  200 2 0.07
>> [log in to unmask] 08 25562.33  200 2 0.07
>> [log in to unmask] 08 25562.33  200 2 0.07
>> ...
>>
>> Thanks for your help
>> Aurelie
>>
>>
>> Le 29/05/13 15:56, Sjors Scheres a écrit :
>>> Hi Aurelie,
>>>
>>> Some things to check:
>>> Is the initial 3D reference map of the correct size?
>>> Were the images normalized?
>>> Are there any all-zero images in your data set?
>>>
>>> If that's all OK, please send the relion command. Also, please share 
>>> your solution on ccp-em once you've found the solution.
>>>
>>> Hope that helps,
>>> Sjors
>>> On 05/29/2013 02:52 PM, Aurélie Nicolas wrote:
>>>> Hello,
>>>>
>>>> I have always the same problem (MultidimArray shape:  Empty 
>>>> MultidimArray!) when I run a 3D classification.
>>>>
>>>> My input star file for relion_refine seems good, it was written by 
>>>> relion_preprocess and looks the same as the star file I used to run 
>>>> your example : ribosomes wi/wo EFG (links to files in EMDB in the 
>>>> wiki are dead but in Google we can access).
>>>>
>>>> The first lines of my input star file :
>>>> data_images
>>>>
>>>> loop_
>>>> _rlnImageName #1
>>>> _rlnMicrographName #2
>>>> _rlnDefocusU #3
>>>> _rlnVoltage #4
>>>> _rlnSphericalAberration #5
>>>> _rlnAmplitudeContrast #6
>>>> [log in to unmask]          01 19999.540000   200.000000 
>>>> 2.000000     0.070000
>>>> [log in to unmask]          01 19999.540000   200.000000 
>>>> 2.000000     0.070000
>>>>
>>>> I think problem appears in it000_model.star where, for each 
>>>> data_model_class#, _rlnSsnrMap #4 is "nan" as you can see :
>>>>
>>>> data_model_class_1
>>>>
>>>> loop_
>>>> _rlnSpectralIndex #1
>>>> _rlnResolution #2
>>>> _rlnAngstromResolution #3
>>>> _rlnSsnrMap #4
>>>> _rlnGoldStandardFsc #5
>>>> _rlnReferenceSigma2 #6
>>>> _rlnReferenceTau2 #7
>>>>             0     0.000000   999.000000          nan 0.000000     
>>>> 0.000000 9.844030e-04
>>>>             1     0.002149   465.407990          nan 0.000000     
>>>> 0.000000     0.036800
>>>>             2     0.004297   232.703995          nan 0.000000     
>>>> 0.000000     0.095343
>>>>
>>>> and all data_model_group# data blocks are empty.
>>>>
>>>> How this snrMap value is calculated ? This is the problem or a 
>>>> consequence of the error?
>>>> Thanks
>>>>
>>>>
>>>> Le 29/04/13 18:13, Sjors Scheres a écrit :
>>>>> Probably your star file header is not OK. Carefully check the header
>>>>> labels and whether the images can be read from the directory from 
>>>>> which
>>>>> you run relion.
>>>>> Hope that helps, Sjors
>>>>>> Hello,
>>>>>>
>>>>>> I try 3d classification of 7450 particles and have this error at the
>>>>>> beginning of the process :
>>>>>>
>>>>>>    Check for dimension: 1
>>>>>> MultidimArray shape:  Empty MultidimArray!
>>>>>> Check called from file ./src/multidim_array.h line 3329
>>>>>>
>>>>>> My command line is :
>>>>>>
>>>>>> mpirun -np 24 -machinefile hosts.mpi
>>>>>> /opt/relion-1.1/bin/relion_refine_mpi --o ex_refine --i ex_norm.star
>>>>>> --particle_diameter 90 --angpix 3.636 --ref 3D-3B.mrc
>>>>>> --firstiter_cc --ini_high 100 --iter 25 --tau2_fudge 4 
>>>>>> --flatten_solvent
>>>>>> --zero_mask --ctf
>>>>>> --ctf_corrected_ref --ctf_phase_flipped --sym C1 --K 2 
>>>>>> --oversampling 1
>>>>>> --healpix_order 2
>>>>>> --offset_range 6 --offset_step 2 --dont_check_norm --norm --scale
>>>>>> --random_seed 1 --j 8
>>>>>>
>>>>>> Previously, I normalised my particles CTF-corrected in RELION and 
>>>>>> have
>>>>>> no errors.
>>>>>>
>>>>>> mpirun -np 24 -machinefile hosts.mpi
>>>>>> /opt/relion-1.1/bin/relion_preprocess_mpi --oper
>>>>>> ate_on all_images.star --o ex_norm --norm --bg_radius 90
>>>>>>
>>>>>> What is the problem ?
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>> -- 
>>>>>> Aurélie NICOLAS
>>>>>> UMR CNRS 6290 - IGDR
>>>>>> Equipe TAF "Translation And Folding"
>>>>>> Bât. 13 - Campus de Beaulieu
>>>>>> 35042 RENNES Cedex - FRANCE
>>>>>> Tel: (33)2 23 23 46 18
>>>>>>
>>>>
>


-- 
Aurélie NICOLAS
UMR CNRS 6290 - IGDR
Equipe TAF "Translation And Folding"
Bât. 13 - Campus de Beaulieu
35042 RENNES Cedex
Tel: (33)2 23 23 46 18