REF: main workflow and utilities #8
Conversation
dPys
force-pushed
the
nipreps
branch
3 times, most recently
from
2cc3974 to
404499a
Compare
|
Latest wf (estimates sigma on the raw data, but saves it for post-interpolation (i.e. eddy) to use for denoising): |
|
TODO: Reduce size of Docker image (currently huge) by removing fslpy stuff from the 6.0.1 source install. |
oesteban
force-pushed
the
master
branch
2 times, most recently
from
c4f0cf4 to
662f586
Compare
dPys
changed the title
|
Hey @dPys, do you think you'd be able to split this PR in more targeted, modular PRs? I can help with the git part if you want. |
|
Yep, I had a feeling that would probably need to happen. Just FYI, this will produce a ton of PR's (you can probably tell this from the commit history within this PR). What do you think about the following approach:
It would be great if you could handle the git part? Let me know what you think about the above strategy and how you think it would be best to go about this. |
|
Sounds great! For most the git part, I believe it could be done with simple |
dPys
force-pushed
the
nipreps
branch
6 times, most recently
from
9536e88 to
98a43d6
Compare
doc: misc setup and doc changes maint: remove unnecessary files Some reorganization, and rethinking the main workflow. Still not runnable via argparse. fix eddy_quad bindings update fork remove slice timing req update Dockerfile fix imports in qc update MANIFEST.in update bids.py update bids.py update Dockerfile to include new eddy turn off cnr_maps with eddy update Dockerfile with fsl 5.0.11 add nlsam denoising to dockerfile add nlsam denoising option add nlsam denoising option fix nlsam, add eddy args fix nlsam, add eddy args create new cli skeleton create new cli skeleton create new cli skeleton add support for multiple sessions and multi-file concatenation add meta-wf handling reorder workflow stages in light of recommendations from Maximov et al. 2019, Ades-Aron et al. 2018(a), Ades-Aron et al. 2018(b) rename final dwi output to make it identifiable adjust ART norm thresh split noise estimation stage and denoising tweak cores/mem alloc liter eddy fix verbosity, refine eddy memory reqs save final bval/bvec as txt update Dockerfile, reduce size fix bug with eddy_quad base_name workflow end-end completion update Dockerfile tweak mem restrictions update mem restrictions fine-tune resource handling fine-tune resource consumption more resource handling fine-tuning refine resource handling add support to drop outliers from sigma if 4d openMP handling squash/merge final pre-meeting add sigma switch to drop_outliers_fn fix sigma fix sigma issue change os.rename to shutil.copy to handle linking across mounted filesystems resource handling remove array passing to alleviate memory/pickling errors update eddy mem reqs fix topup odd slice option update eddy mem reqs fix b02b0.cfg bug for odd slices simplify b02b0.cnf specify handling for second-level modeling in eddy if data is detected as hemispheric specify handling for second-level modeling in eddy if data is detected as hemispheric FIX: send empty str to 'args' trait of an EddyInputSpec empty str to eddy args for non-reverse-phase empty str to eddy args for non-reverse-phase: PR fix slm str trait bug adjust .cnf hyperparameter window adjust .cnf hyperparameter window
|
Hello @dPys, Thank you for updating!
To test for issues locally, Comment last updated at 2019-11-04 16:34:47 UTC |
|
Hey @dPys : what of all this (if anything) is still relevant. Looks like a lot of the commits are coming from elswhere? We can look at cherry picking tomorrow, if that's still something we need to do. |
|
Hi @arokem -- let's go ahead and close this PR since most of this is either being incorporated through separate PR's already or will be! |
|
Okidoke! |
This is my attempt at slimming things down and consolidating. Others' WIPs can still be added back in, but I wanted to PR something bare-bones to start.