dcbia-ortholab / mfsda_python Goto Github PK

Currently only showing Loic Michoud.

Will update to @vicory, @jcfr, @MahmoudMostapha as well as Loic.

There is a tutorial powerpoint dangling in #10 from Mahmoud Mostapha, is there a DCBIA google drive where I could upload it as Google slides ?

Or since it corresponds to a tutorial written for SlicerSALT, is there already a public URL for it ?

Motivation: This tutorial would be referenced in the README and help user understand how to use the extension.

Challenge: Because of naming different between the loadable module in SlicerSALT and the default name in the extension, user will likely be confused by reading the tutorial. See #16

I've occasionally run into the code failing when trying to invert a singular matrix. Code could check for this error and use a pseudo-inverse instead.

Would it make sense to promote standalone re-use of the functionality ?

The following would then be possible:

pip install mfsda

The following CLIs are independent of Slicer:

• MFSDA_run.py
• MFSDA_createShapes.py
• MFSDA_selectVariablesPCA.py

The following modules are independent of Slicer:

• MFSDA_stat.py
• stat_*.py

See attached file.

Slicer recently upgraded from Python 3.6.7 to Python 3.9.10 in Slicer/Slicer@34e48e8. There are now observed build errors of the MFSDA Slicer extension as seen at https://slicer.cdash.org/viewBuildError.php?buildid=2581568. This is because the python requirements file in the repo is including hashes that are only for Python 3.6 wheels. These need to be updated for Python 3.9 wheels which also means the actual version of packages here will need to be updated as some older versions simply don't have Python 3.9 wheels because that version was released before Python 3.9 was even released.

master...slicersalt:slicersalt-2019-02-13-0a67df1d8

cc: @juanprietob

Could you associate the 3d-slicer-extension GitHub topic so that it is listed here: https://github.com/topics/3d-slicer-extension

To learn more about topic, read https://help.github.com/en/articles/about-topics

To support distribution of MFSDA as an extension, we would need one or two screenshots.

Current code makes assumption that the covariates read from the file has 2d shape (i.e. rows and columns) while the numpy reader will collapse the second dimension if there is only 1 column.

@chaohstat To avoid confusion, could you delete your project chaohstat/MFSDA_Python and re-fork the one hosted in the DCBIA-OrthoLab organization. See https://github.com/DCBIA-OrthoLab/MFSDA_Python

The fork graph currently shows this:

@AGrigis, @BIG-S2: Could you do the same ?

Thanks 🙏