This plugin provides wrappers for several programs of RELION software suite.
- If you have imported movies with a gain file in DM4 format, you need to flip the gain reference upside-down in the motion correction protocol! (bug details)
- If you have provided a gain reference or defects file during movie import or motion correction and plan to run bayesian polishing, please make sure to run first "assign optics groups" protocol for aligned movies, specifying the gain file etc. Currently, Scipion has no other way of knowing if you have e.g. rotated the gain during motion correction. Output movies then can be used in the polishing protocol.
- When importing EER movies, you should specify dose per single EER frame during import step. Pixel size should be the physical pixel size unless you plan to render EER on 8K grid. In the Motioncor protocol choose binning, EER fractionation and upsampling (default=1 is for 4K grid).
You will need to use 3.0+ version of Scipion to be able to run these protocols. To install the plugin, you have two options:
scipion installp -p scipion-em-relion
- download repository
git clone -b devel https://github.com/scipion-em/scipion-em-relion.git
scipion installp -p /path/to/scipion-em-relion --devel
- RELION sources will be downloaded and compiled automatically with the plugin, but you can also link an existing installation. Default installation path assumed is
software/em/relion-4.0, if you want to change it, set RELION_HOME in
scipion.conffile to the folder where the RELION is installed.
- If you need to use CUDA different from the one used during Scipion installation (defined by CUDA_LIB), you can add RELION_CUDA_LIB variable to the config file. Optionally, you can also specify RELION_CUDA_BIN path for nvcc.
- If you have to use a MPI for Relion different from Scipion MPI, you can set RELION_MPI_BIN and RELION_MPI_LIB variables in the config file.
- If you want to use 2D class ranker protocol, you need to set RELION_ENV_ACTIVATION that points to a conda environment which includes torch and numpy modules. Default =
conda activate relion-python
- If you want to use SIDESPLITTER, you need the sidesplitter plugin installed and SIDESPLITTER_HOME set properly. After that, you only need to add --external_reconstruct into Additional arguments field of a 3D auto-refine or multi-body job.
To check the installation, simply run one of the tests. A complete list of tests can be displayed by executing
scipion test --show --grep relion
- 2D class ranker
- 2D classification
- 3D auto-refine
- 3D classification
- 3D initial model
- 3D multi-body
- assign optics groups
- auto-picking (reference-based)
- auto-picking LoG
- bayesian polishing
- calculate fsc
- center averages
- clean project
- compress movies
- create 3d mask
- crop / resize volumes
- ctf refinement
- estimate gain to compress
- expand symmetry
- export coordinates
- export ctf
- export particles
- import coordinates
- local resolution
- motion correction
- particles extraction
- preprocess particles
- remove preferential views
- subtract projection
- symmetrize volume
- Scheres et al., JMB, 2012
- Scheres et al., JSB, 2012
- Kimanius et al., eLife, 2016
- Zivanov et al., eLife, 2018
- Kimanius et al., Biochemical Journal, 2021