🐛 Spurious condition: (ref_resolution ≟ resolution_for_anat) ≠ is_monkey

[shnizzedy]

Related problem

This conditional block

C-PAC/CPAC/registration/registration.py

Lines 2309 to 2325 in 61ad414

	if (
	cfg.registration_workflows["anatomical_registration"]["registration"][
	"FSL-FNIRT"
	]["ref_resolution"]
	== cfg.registration_workflows["anatomical_registration"]["resolution_for_anat"]
	):
	node, out = strat_pool.get_data("T1w-brain-template")
	wf.connect(node, out, fsl, "inputspec.reference_brain")
	node, out = strat_pool.get_data("T1w-template")
	wf.connect(node, out, fsl, "inputspec.reference_head")
	else:
	node, out = strat_pool.get_data("FNIRT-T1w-brain-template")
	wf.connect(node, out, fsl, "inputspec.reference_brain")
	node, out = strat_pool.get_data("FNIRT-T1w-template")
	wf.connect(node, out, fsl, "inputspec.reference_head")

seems to be conditioned on something spurious (

C-PAC/CPAC/registration/registration.py

Lines 2310 to 2313 in 61ad414

	cfg.registration_workflows["anatomical_registration"]["registration"][
	"FSL-FNIRT"
	]["ref_resolution"]
	== cfg.registration_workflows["anatomical_registration"]["resolution_for_anat"]

) to what the condition should actually be (

C-PAC/CPAC/resources/configs/pipeline_config_default.yml

Lines 725 to 726 in 61ad414

	# Template to be used during registration.
	# It is for monkey pipeline specifically.

)

Proposed feature

Update the condition to reflect the actual condition we should be checking, rather than a condition that probably corresponds to that condition.

Acceptance criteria

• It's obvious which condition will be selected and why.

[shnizzedy]

I think we want to add something like

pipeline_setup:
  # options: human, non-human primate, rodent
  organism: human

to the pipeline configuration and update

C-PAC/CPAC/registration/registration.py

Lines 2309 to 2314 in 61ad414

	if (
	cfg.registration_workflows["anatomical_registration"]["registration"][
	"FSL-FNIRT"
	]["ref_resolution"]
	== cfg.registration_workflows["anatomical_registration"]["resolution_for_anat"]
	):

to

    if cfg.pipeline_setup["organism"] != ["non-human primate"]:

, but I'm unsure:

• if organism is the key we want,
• if human, non-human primate, rodent is the controlled vocabulary we want or if we want to
  • also differentiate between children and adults (in humans and/or in other organisms), and/or
  • differentiate nhps and rodents more specifically than that, or
• if we really need this information or if there's a simpler approach or refactor of the config to achieve the same clarity, or
• what other places we want to use this option for a conditional instead of whatever's already there.

Also, the pipeline validation schema can be updated to make these

C-PAC/CPAC/pipeline/schema.py

Lines 710 to 711 in 61ad414

	"FNIRT_T1w_brain_template": Maybe(str),
	"FNIRT_T1w_template": Maybe(str),

required for nhp pipelines (and these

C-PAC/CPAC/pipeline/schema.py

Lines 694 to 695 in 61ad414

	"T1w_brain_template": Maybe(str),
	"T1w_template": Maybe(str),

might be optional for nhp pipelines? Or maybe all we need to do is get rid of the FNIRT_T1w* keys altogether and do

registration_workflows:
  anatomical_registration:
    T1w_brain_template: /cpac_templates/MacaqueYerkes19_T1w_1.0mm_brain.nii.gz
    T1w_template: /cpac_templates/MacaqueYerkes19_T1w_1.0mm.nii.gz

in the nhp pipelines?)