Thank you for the response! I'm an idiot and sent you star files... attached now are screenshots of representative particles and early classes. Also, two images showing some of the worst background variation in the orig micrographs.
The micrographs show more than just simple gradients: 1) flatfields weren't always working during collection so you can also see distinct top and bottom zones in addition to general left/right gradient, and 2) the second half of the data include BLACK gold fiducials. Guessing the latter could be addressed through the --black_dust option. I may be out of luck with the former -- it's probably not a common problem.
Most basically, I'd like to know whether, for a properly done normalization, I should expect to see none of that variation I'm seeing in the particles now. I can't recall ever noticing it in earlier datasets.
And, yes, had since figured how to run mpi version (set num MPI procs!). I'm using SLURM job handler and learning :-) Thank you again for any thoughts!
Best,
Nancy
Nancy Meyer
Research Associate
Oregon Health and Science University
Department of Biochemistry and Molecular Biology
Chapman Lab, MRB 534
Ph: 503-494-1615
Fx: 503-494-8393
________________________________________
From: Sjors Scheres [[log in to unmask]]
Sent: Wednesday, January 27, 2016 12:05 AM
To: Nancy Meyer
Cc: [log in to unmask]
Subject: Re: [ccpem] Adequate particle normalization prior to 2D classification?
Hi Nancy,
In 1.4, normalization should include ramping background correction. Are
these images which you send? What format are they in? I cannot seem to
open them.
best,
Sjors
PS: shouldn't you run relion_preprocess_mpi instead of relion_preprocess?
(this will not solve your issue)
> Hi all,
>
> I'm concerned about whether a particle set is being normalized properly
> during particle extraction in Relion 1.4. Attached is a star file of
> representative particles (icosahedral virus). At the bottom is my particle
> extraction command.
>
> Backgrounds seem to vary quite a bit, even when boxes are not over the
> edge of the micrograph. I should say a good number of the micrographs show
> a regular light/dark gradient across the image (due to acquisition
> issues), and that each contain at least a single row of "dead" pixels.
> Particles were extracted from raw micrographs following CTF estimation in
> Relion using .box coord files determined in EMAN2, so these images should
> not have any normalization done to them except that specified by Relion's
> --norm option.
>
> Concerned that intensity variation will be problematic downstream -- can
> anyone suggest Relion-compatible methods to take into account the
> background gradients (if that's the culprit), or am I doing something more
> obviously wrong? Thanks for any input!
>
> Command for particle extraction:
> -bash-4.1$ srun --nodes 5 --ntasks-per-node 24 relion_preprocess --o ptcls
> --mic_star micrographs_ctf.star --coord_suffix ".box" --extract
> --extract_size 384 --norm --bg_radius 190 --white_dust -1 --black_dust -1
> --coord_files "Micrographs/*.box"
>
> -- particles themselves are approximately 260A diam
> -- also attached an example of early 2D classes showing background
> variation
>
> Thanks!!
>
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
|