https://github.com/diamondlightsource/python-relion-yolo-it
EM pipeline relion_it with crYOLO support
https://github.com/diamondlightsource/python-relion-yolo-it
Last synced: 3 months ago
JSON representation
EM pipeline relion_it with crYOLO support
- Host: GitHub
- URL: https://github.com/diamondlightsource/python-relion-yolo-it
- Owner: DiamondLightSource
- License: gpl-2.0
- Created: 2019-09-18T15:28:46.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2021-03-30T15:44:35.000Z (about 4 years ago)
- Last Synced: 2025-02-07T22:05:08.439Z (3 months ago)
- Language: Python
- Homepage: https://pypi.org/project/relion-yolo-it/
- Size: 1.34 MB
- Stars: 5
- Watchers: 8
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# relion_it with crYOLO support
In options you can choose to pick via the crYOLO
general model or with the relion auto picker. CrYOLO runs as an external job after ctfFind. Relion
then takes the particle coordinates found by crYOLO and then further processes them. Particles
appear as a manual pick job in the relion gui and can be viewed there.
## Requirements for external use:
1. CrYOLO and Relion 3.0 installed.
2. Conda Environment for crYOLO and Relion
3. Edit paths in options.py for MotionCor2 and Cryolo general model locations
4. Run by: `cryolo_relion_it.py /Path/To/options.py --gui`
## Scripts being used:
- `cryolo_relion_it.py`: The main script called by `dls_yolo_relion`. This houses the main pipeline
and calls to all the other scripts.
- `cryolo_pipeline.py`: The crYOLO pipeline. This runs as a subprocess and exectutes many repeated
times to Import, MotionCorr, CtfFind, crYOLO pick, Extract... as new movies are
collected. As Relion 3.0 does not support external job types the YOLO pipeline is in fact 3
seperate pipelines chained together.
- `cryolo_external_job.py`: Reads Relion star files and makes a directory that crYOLO can execute
particle picking from.
- `cryolo_fine_tune_job.py`: After 2D classification, good classes can be selected to fine tune the
cryolo general model. After the finetuning, crYOLO uses this new model to pick future
particles in the current run.
- `dls_options.py`: Example template options for the installation at DLS / eBIC.
Note: Fine-tuning requires good 2D classes to be picked by hand after first 2D iteration
## Contributors
* Sjors H.W. Scheres
* Takanori Nakane
* Colin M. Palmer
* Donovan Webb