carpentries-incubator / sdc-bids-fmri Goto Github PK

Currently this lesson is rendering with a Data Carpentry logo. @fmichonneau - could you please pull in the remote theme and update to Incubator branding?

With the requirement of Carpentries to use their markdown format and our preferred live teaching medium being Juypter Notebooks we've run into an issue of having to update the Carpentries MD and Jupyter Notebooks independently anytime an update is made to the lessons.

Tools like https://github.com/mwouts/jupytext enable automated Markdown/Notebook/Script generation from a single source file. However, an issue that we're running into is with the generation of data carpentry markdown tags.

This issue is to discuss ways in which we can integrate the data carpentry markdown tags into our future source format (possibly Jupytext MyST?)

Introduce the changes in #39 into the notebooks.

This is an on-going issue for completing of the beta phase of this workshop. Part of the requirement is the development of an instructor guide which orients new carpentries instructors to the material flow, teaching medium, as well as common technical issues that pop up during teaching.

This documentation should be placed under _extras/guide.md.

For example, sometimes (ep 6) it's imported as:

from nilearn import plotting

where others (ep 3, ep 4, ep 7) do:

from nilearn import plotting as plot

In one case (ep 5), it's aliased as plt but then matplotlib.pyplot uses plt too.

With the continuing development of nilearn statistics (formerly nistats) it is a good time to consider building out a task GLM module. Some work towards task fMRI has already been done here #9, but it needs to be updated to reflect the merging of nilearn and nistats.

This lesson is currently rendering with a "pre-alpha" banner. From quickly skimming through the materials, however, it looks pretty complete. I would recommend changing the lifecycle tag in _config.yml to "alpha".

Could you provide me access to some part of the entire dataset that you have? I wish to perform some experiments and want to be able to generate correlation maps between various parcels/voxels. Having access to a few subjects would be splendid!

The current dataset used in the lessons is running a (very) legacy version of fMRIPrep (0.4.0). Since then, fMRIPrep has changed its output structure to follow the BIDS derivatives convention more closely and as a result the outputs learners are dealing with during these lessons do not map to what they would see when running their own analyses.

Ideally we'd use pre-existing fMRIPrep 20.2.0 LTS outputs, however with the requirement that the dataset must have a case-control comparison with resting state data and available task data we've decided to run the subset of individuals we use in this lesson through LTS and host the data on OSF.

Instead of using the AWS S3 cli to access data, we will switch over to using Datalad to follow more closely to SDC-BIDS-IntroMRI; the pre-requisite to this course.

TODO:

• Set up fMRIPrep 20.2.3 LTS run
• Run data through fMRIPrep 20.2.3 LTS
• Host data on OSF
• #31
• Deprecate setup scripts that rely on AWS CLI
• Update Data Carpentry markdown to reflect changes in Jupyter