neurodata / m2g

NeuroData's MRI to Graphs (m2g) - connectome estimation package and pipeline
https://docs.neurodata.io/m2g/
Other
64 stars 36 forks source link

[WIP] Bunch of bug fixes and features #398

Closed j1c closed 2 years ago

j1c commented 4 years ago
  1. ArgumentParser import is missing in m2g_dwi_pipeline, causing an error in case someone wants to run dwi_pipeline separately
  2. fix variable used before assignment error in plot_overlays
  3. fix default file names for mask files in DmriReg
  4. add proper file checking for gen_tissue output when using --skipreg
  5. Limit fury version. Latest versions do not work due to deprecations. Limited to the version in m2g:latest docker image.
    • Just disabled this code entirely as its causing issues when not in docker. Its broken and and kills the process.
  6. black on some files
  7. add check for parcellations. too many times i ran pipeline runs but realized there were no parcellations to begin with. maybe add a list of valid parcellations.
  8. add multiprocessing to connectome generation. there were lots of inefficiencies to begin with. this reduces time by A LOT.
  9. Moved the argument parsing outside of main() in m2g_bids.py so that arguments can be passed when not called as a script, but imported in different package.
  10. Parallelized tracking via splitting the seeds into chunks. This is simply a parallelization of a for loop. Proof here: notebook
Lawreros commented 3 years ago

I've merged your changes with the branch I've been working on cpac-py3, currently going through test to make sure everything runs ok. I think I may have accidentally changed your branch as well as a result of correcting the commit conflicts, so feel free to undo that commit