Hi Everyone,
I am new to the list-serv but I have found everyone's questions and answers incredibly helpful!
I am processing cryo-em data which has been motion-corrected with dose-weighting in Leginon then imported and pre-processed the rest of the way in Relion v2.0.5. After auto-picking, I 4-times binned (box size 400 to 100) the data using the particle extraction tab. Unfortunately, when I tried to run this through 2D classification I got the following error:
fn_img= [log in to unmask] bg_avg= -0.959901 bg_stddev= 1.76119 bg_radius= 104.944
ERROR:
ERROR: It appears that these images have not been normalised to an average background value of 0 and a stddev value of 1.
Note that the average and stddev values for the background are calculated:
(1) for single particles: outside a circle with the particle diameter
(2) for helical segments: outside a cylinder (tube) with the helical tube diameter
You can use the relion_preprocess program to normalise your images
If you are sure you have normalised the images correctly (also see the RELION Wiki), you can switch off this error message using the --dont_check_norm command line option
File: /softlib/apps/EL7/relion/2.0.4/relion_benchmarking/src/ml_optimiser.cpp line: 1713
At the prompt of the error and looking around this forum, I found that this can sometimes happen when re-scaling images during extraction. So I added the --dont_check_norm flag. This fixed the issue and the output looked pretty good at iteration 29 (there were classes with well defined structure for my molecule). I stopped the job at this point but later decided that it was not yet converged.
Unfortunately, when I tried to continue the run from iteration 28 using the following command:
`which relion_refine` --continue Class2D/job025/run_it028_optimiser.star --o Class2D/job025/run_ct28 --dont_combine_weights_via_disc --preread_images --pool 16 --iter 100 --tau2_fudge 2 --particle_diameter 450 --oversampling 1 --psi_step 12 --offset_range 5 --offset_step 2 --j 8 --gpu "" --dont_check_norm
I get the following warning and error:
The following warnings were encountered upon command-line parsing:
WARNING: Option --dont_check_norm is not a valid RELION argument
ERROR:
Array_by_array: different shapes (*)
File: /softlib/apps/EL7/relion/2.0.4/relion_benchmarking/src/multidim_array.h line: 2899
So it seems to me that relion_refine with the --continue flag does not recognize the --dont_check_norm option. This leaves me in the predicament that I cannot continue any of these 2D Classifications.
Is there any work-around? Thank you!
Ambrose Plante
|