🧑‍💻 Redesign boxplot setup

[shnizzedy]

Planning

No response

User interface changes

No response

Project sub-parts and estimates

• combine into fewer charts?
• clarify axis labels

Timeline notes

No response

Timesink notes

No response

Frustration notes

No response

Design decisions and notes

• currently it spits out a bunch of separate files like (this example is just one subject. They're box-and-whisker plots when there are more subs)

[sgiavasis]

One thing I'm picturing is a single main "quick view" boxplot where outputs are included roughly in order of when they were produced. This would make it easier for the developers to eyeball where a variation between pipelines was introduced. This doesn't necessarily need to replace the existing boxplots which are sorted by data type and can just be an addition.

Example of the outputs in this one figure:

• desc-preproc_T1w
• desc-brain_mask
• label-CSF_mask
• label-WM_mask
• label-CSF_desc-preproc_mask
• label-WM_desc-preproc_mask
• space-{template}_desc-preproc_T1w
• desc-mean_bold
• sbref
• space-T1w_sbref
• space-{template}_sbref
• desc-preproc_bold - NIfTI time series
• desc-confounds_timeseries - TSV columns with labels
• space-{template}_desc-preproc_bold - NIfTI time series

Important: Depending on the pipeline configuration, some of these outputs may not be present, so the boxplot script should be able to be flexible regarding this, while also catching when any of these outputs are supposed to be present, but are missing (or if the filepath matching failed).

[sgiavasis]

It would also be nice to separate out the participant data that use distortion/susceptibility correction into their own three sets of boxplots (but still keep the total conglomerate ones also):

• PhaseDiff field maps
• EPI field maps
• EPI field maps but for abcd-options and any other config which uses FSL TOPUP instead of AFNI 3dQwarp

[sgiavasis]

@birajstha - from earlier conversation, check out @e-kenneally 's issues with the template-space preprocessed BOLD time series correlations not completing/showing. This is a good focus for a unit test as well.

[birajstha]

@birajstha - from earlier conversation, check out @e-kenneally 's issues with the template-space preprocessed BOLD time series correlations not completing/showing. This is a good focus for a unit test as well.

This issue has been resolved, will resume the original task.

[sgiavasis]

@birajstha I did notice they're also missing in the abcd-options plots - see if that's a separate issue?

[sgiavasis]

Adding another potential feature idea - it would also be nice if there was an easy interface to grab the individual correlations of each output for each participant for each config. Without having to dig.

Ex. "Can I see the correlation of the space-MNI152NLin6Asym_desc-preproc_bold for participant sub-ETC for the rbc-options config?"

[birajstha]

@birajstha I did notice they're also missing in the abcd-options plots - see if that's a separate issue?

File paths matching were failing causing this missing plot. They have been resolved by adding replacements in the YML file.