add models for SPM block design tutos

[Remi-Gau]

replaces #1

TODO

• Make it work with fitlins

[shashankbansal6]

I believe we moved type into Model
Something like this: "Model": {"X": [1], "Type": "Meta"},
Ref:

model-zoo/narps/model/model-narps_smdl.json

Line 48 in 6b38841

"Model": {"X": [1], "Type": "Meta"},

[Remi-Gau]

I believe we moved type into Model

Oh gosh. That JSON has like 10 lines and I managed to F it up. Thanks for catching this.

[Remi-Gau]

Currently this fails to run with fitlins:

$ bash moae_run_fitlins.sh

Captured warning (<class 'FutureWarning'>): Fetchers from the nilearn.datasets module will be updated in version 0.9 to return python strings instead of bytes and Pandas dataframes instead of Numpy arrays.
Traceback (most recent call last):
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/bin/fitlins", line 8, in <module>
    sys.exit(main())
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/fitlins/cli/run.py", line 432, in main
    sys.exit(run_fitlins(sys.argv[1:]))
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/fitlins/cli/run.py", line 385, in run_fitlins
    fitlins_wf = init_fitlins_wf(
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/fitlins/workflows/base.py", line 285, in init_fitlins_wf
    select_specs = pe.Node(
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/nipype/pipeline/engine/nodes.py", line 181, in __init__
    super(Node, self).__init__(name, kwargs.get("base_dir"))
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/nipype/pipeline/engine/base.py", line 37, in __init__
    self.name = name
  File "/home/remi/github/BIDS-model-zoo/spm_tutorials/env/lib/python3.8/site-packages/nipype/pipeline/engine/base.py", line 53, in name
    raise ValueError('[Workflow|Node] name "%s" is not valid.' % name)
ValueError: [Workflow|Node] name "select_run level_specs" is not valid.

[adelavega]

Hey, this is really nice! I like how reproducible it all is.

A more general concern I have is that I'm hoping that all model zoo examples eventually follow a similar format, so that they can eventually be displayed in a Jupyter Book or equivalent website, and run in Binder.

I'll probably hold off on merging until we have that, but this looks very compatible.

[Remi-Gau]

Well some of the stuff for fitlins are also there because I wanted to make it easier for others to run it and help me figure out what is wrong with this model and why it can't run.

[shashankbansal6]

Finally, after all of this fixes, it should run the FirstLevelModel and you will get all of the first level images but I just discovered a bug in fitlins reports where its hard-coded to expect a subject level node so it will throw the following error:

Traceback (most recent call last):
  File "/Users/shashankbansal/miniconda/envs/spm_tuto/bin/fitlins", line 8, in <module>
    sys.exit(main())
  File "/Users/shashankbansal/miniconda/envs/spm_tuto/lib/python3.8/site-packages/fitlins/cli/run.py", line 432, in main
    sys.exit(run_fitlins(sys.argv[1:]))
  File "/Users/shashankbansal/miniconda/envs/spm_tuto/lib/python3.8/site-packages/fitlins/cli/run.py", line 425, in run_fitlins
    report_dict = build_report_dict(deriv_dir, work_dir, graph)
  File "/Users/shashankbansal/miniconda/envs/spm_tuto/lib/python3.8/site-packages/fitlins/viz/reports.py", line 130, in build_report_dict
    {analysis_dict['entities']['subject'] for analysis_dict in report['nodes'][0]['analyses']}
  File "/Users/shashankbansal/miniconda/envs/spm_tuto/lib/python3.8/site-packages/fitlins/viz/reports.py", line 130, in <setcomp>
    {analysis_dict['entities']['subject'] for analysis_dict in report['nodes'][0]['analyses']}
KeyError: 'subject'

I have created a new issue in fitlins to keep track.

[adelavega]

Should we merge and later change things to a the uniform environment once that is better defined?

[Remi-Gau]

Should we merge and later change things to a the uniform environment once that is better defined?

good with me

this may also include some change, once this is runnable with fitlins