🐛 `TypeError: string indices must be integers` when forking on `space: ['native', 'template']`

[shnizzedy]

Describe the bug

Traceback (most recent call last):
  File "/code/CPAC/pipeline/cpac_pipeline.py", line 445, in run_workflow
    subject_id, sub_dict, c, p_name, num_ants_cores
  File "/code/CPAC/pipeline/cpac_pipeline.py", line 1464, in build_workflow
    wf = connect_pipeline(wf, cfg, rpool, pipeline_blocks)
  File "/code/CPAC/pipeline/cpac_pipeline.py", line 1102, in connect_pipeline
    wf = nb.connect_block(wf, cfg, rpool)
  File "/code/CPAC/pipeline/engine.py", line 1426, in connect_block
    pipe_x, opt)
  File "/code/CPAC/nuisance/nuisance.py", line 2678, in nuisance_regression_template
    'template')
  File "/code/CPAC/nuisance/nuisance.py", line 2475, in nuisance_regression
    if res is None else
TypeError: string indices must be integers

To reproduce

Run C-PAC 1.8.5 with

nuisance_corrections:
  2-nuisance_regression:
    space: ['native', 'template']

Preconfig

• [ ] default
• [ ] abcd-options
• [ ] anat-only
• [ ] blank
• [ ] ccs-options
• [ ] fmriprep-options
• [ ] fx-options
• [ ] monkey
• [ ] monkey-ABCD
• [ ] ndmg
• [ ] nhp-macaque
• [ ] preproc
• [ ] rbc-options
• [ ] rodent

Custom pipeline configuration

nuisance_corrections:
  2-nuisance_regression:
    space: ['native', 'template']

Run command

No response

Expected behavior

A double-set of outputs of nuisance regression and everything downstream

Acceptance criteria

• [ ] C-PAC does not crash when forking on nuisance_corrections, 2-nuisance_regression, space

Screenshots

No response

C-PAC version

v1.8.5

Container platform

No response

Docker and/or Singularity version(s)

No response

Additional context

The root cause is that regressors (in 1.8.5, renamed to desc-confounds_timeseries in 1.8.6) is both an input and an output here (the nuisance_regression_native NodeBlock), so when template-space tries to run, it wants to run on the outputs of native-space-regression. I think there’s probably a few pretty graceful solutions here, but I don’t know enough about nuisance regression to know how we’d want to tackle it.

1. Depending on whether we’re generating the regressors only or generating and running, we want the outputs to be named the same either way? (I believe that’s the current behavior)
2. When we run nuisance regression, we don’t need the generated-not-run version anymore? (I believe it gets overwritten currently)
3. I’m not sure how this interacts with ingressing regressors

---
title: "Resource: 'regressors'"
---
flowchart TD
subgraph "generate only"
    nuisance_regressors_generation
end
subgraph "one space"
    nuisance_regressors_generation_1[nuisance_regressors_generation] --regressors--> nuisance_regression_1[nuisance_regression]
end
subgraph "forking (intended)"
    nuisance_regressors_generation_2[nuisance_regressors_generation] --regressors--> nuisance_regression_native_2[nuisance_regression_native]
    nuisance_regressors_generation_2 --regressors--> nuisance_regression_template_2[nuisance_regression_template]
end
subgraph "forking (actual)"
    nuisance_regressors_generation_3[nuisance_regressors_generation] --regressors--> nuisance_regression_native_3[nuisance_regression_native] --regressors--> nuisance_regression_template_3[nuisance_regression_template]
end
nuisance_regressors_generation --regressors--> output_directory
nuisance_regression_1 --regressors--> output_directory
nuisance_regression_native_2 --regressors--> output_directory
nuisance_regression_template_2 --regressors--> output_directory
nuisance_regression_template_3 --> crash_typeError[`TypeError: string indices must be integers`]