Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.

Processing using Relion3.0/Motioncorr

Background

Please note that your movies (mrc files written out by the microscope) are currently on the strubi system. This means that the files can only be ‘seen’ when ssh’d into compG000, compG001 and Belmont.

Cryosparc currently runs on compG001 as are all command-line run GPU jobs.

 

The Relion3.0 GUI offers two ways to run motion correction:

a.         Using relion’s own implementation of motion correction. THIS CAN ONLY BE RUN ON CPUs!

b.         Using relion’s wrapper for Motioncorr.

          

First sym link your micrographs to your directory. 

ln –s /data/bowmore/<glacios||krios>/<PI>/<nameofyourmicrographdirectory>/* .

 

Check that this directory exists first. If your sym links appear red, then it is likely that your entered path does not exist.

 

Running motion correction…

 

OPTION B: MOTIONCORR

Please avoid using compG001 at present, since this is used for cryosparc jobs.

ssh -X compg000

module load relion/3.0_beta-gcc5.4.0-cuda9.0-20190207

module unload cuda/9.0

module load cuda/9.1

cd /well/strubi/person/YourDataFolder

relion

In the gui, import files (e.g. Movies/*.tif)

For option b. motioncorr:

Use the following executable: /apps/strubi/motioncorr/2-1.1.0-gcc5.4.0-cuda9.1-sm70/MotionCor2

Tick NO for "Use RELION's own implementation".

Set the GPU ID.

To check which GPUs are unused, on a terminal connected to compg000 enter:

nvidia-smi

Then select ‘run’ on the GUI. It is worth checking using the nvidia-smi command that your job is running where you expect it to be running!

 

OPTION A. RELION MOTIONCORRECTION 

Please do not run option a. motion correction on the gpu cores.

 This can currently only run on cpus, and thus is not appropriate for compG. ssh into Belmont, and submit your job to the cluster.

 

ssh –X someone@belmont

module load relion/3.0_beta-gcc5.4.0-CentOS6-2019020

relion

 

NOTE that the tutorial says that, it is optimal to run this correction on multiple threads, such that each thread can run independently on a movie frame and thus it is best to use an integer of <N movie frames>/<N threads>. This is not currently possible. However, as each thread will work independently on a movie frame, it is optimal to use a number of threads such that the number of movie frames divided by the number threads is an integer number. With movies having 24 frames, using 12 threads will result in 2 frames being processed by each thread.

 

Remaining Relion Pipeline (CTF correction to PostProcessing)

 

The rest of the relion pipeline should be run using the rescomp queue system (example submission script below). [LC1] 

 

Ensure that the most recent version of relion3.0 is sourced using the following command:

module load relion/3.0_beta-gcc5.4.0-cuda9.0-20190207

Dedicated relion nodes can be found on queue C (16 cores per node) and queue E (24 cores per node), and additional highmem nodes may be accessed on request (email rescomp).

 

To check that your job is running, use ‘qstat’. Consult the sungrid engine website (useful documentation below) for more information.

 

For CTF processing, if people want to use GCTF they can do it on compg000 node because it is GPU dependent process and they can do it by using the following executable:

- for phase plate data: /users/strubi/jun/apps/Gctf/Gctf-v1.18_sm_30_cu8.0_x86_64

- for none phase plate data: ~juha/apps/gctf/Gctf-v0.1.06_sm_20_cu7.5_x86_64

 

Running Relion with Warp Coordinates

 

  1. Copy warp coodinates over to your relion working directory. Coordinates should be located in <pathtowarpprocessing>/matching/*.star . To avoid copying over the 'false positives', ensure that the wild card for the star file includes *<pickingname>.star

cd relion-wd

mkdir micrographs

cp /data/bowmore/krios/bowden/vicky/awesome-data/matching/*propicked.star micrographs/.

It is actually cleaner to copy your coordinates over to whichever directory you have sym-linked your micrographs to.

  1. Change the coordinate star file names to adhere to relion's expectations.

cd coordinates

rename propicked _pick *.star

 

  1. Remove spurious headers from warp using custom scripts. Moan to HelenD if this fails.

sh  /well/strubi/hmed/apps/microscopy/convertwarpcoords/runremove.sh

 

  1. This should remove the spurious headers from offending files. To avoid screwing up, some 'old' star files. Move these out of the way into a temporary directory for the time being.

mkdir old

mv old*star old/.

  1. Now you should be ready to import with relion. Open the relion gui, navigate to the import tab, and enter the file path to your coordinates into the appropriate window. In the dropdown menu, select 'Import 2D/3D coordinates'.

coordinates/*_pick.star

  1. Check that the data has been successfully imported by going into the Import directory and corresponding job number. There should be a star file named according to the provided suffix, and a fold with your coordinate directory (in this case 'coordinates') that contains all of your coordinate star files (yes, another copy of them->#relion_efficiency).
  2. At this stage, it is likely that your coordinate path does not correspond to the micrograph path. There are two ways to avoid this. (a) Post-hackfix-emulate the micrograph file directory in your Import/job00x/coordinates i.e. Import/job00x/<path-to-micrographs> . (b) Start from number 1, and copy coordinates over to your micrographs directory. We did (a) first time round.
  3. Run extraction! Open the relion gui, and run extract with your Import/job00x/coordinates_pick.star files and micrographs_ctf.star micrographs. This should should use the dose-weighted micrographs that are listed in the micrographs star file.

 

Relion3.0 Bayesian Particle Polishing

 

Relion needs to be able to see your raw movies to run the Bayesian polishing procedure. Therefore, this job must be queued on either compG000 (if GPU-compatible), or Belmont.

 

To run on Belmont, you must run the following:

module load relion/3.0_beta-gcc5.4.0-CentOS6-2019020

 

Please do not run your polishing job directly on Belmont, and instead run on the queue system (16 cores).

 

Common Problems

 

  1. libcufft.so.9.0 library error.

This is because of an incompatability of library versions, likely because you sourced the most recent version of relion on rescomp1/rescomp2. Instead, either load the earlier version of relion, or ssh into compG000 or compG001, and source the sge script mentioned above.

 

Useful links

Internal pages of the strubi website.

RELION3.0: ftp://ftp.mrc-lmb.cam.ac.uk/pub/scheres/relion30_tutorial.pdf

WARP: www.warpem.com/warp/