How deal with missing data in nipype

1

How do we deal with missing data in nipype? (how can we disconnect a branch of a workflow is file is missing)
I tried to play with ignore_exception at the Node level but with no much success.

Let’s say you have:

  • A study with multiple visit (t0, t1, …) and multiple modalities (mod0, mod1, …),
  • A really simple workflow that consists in co-registering modi(ti) to mod0(t0)
  • For some time points, m1 is missing.

Let’s assume that:

  • folder structure is: /{sub_id}/{visit_id}/{mod_id}
  • each modalities has its own co-reg node.
  • infosource in looping over visit_id
  • select file is defined as: selectfile = Node(SelectFiles({'mod0': '/sub_id/{visit_id}/mod0', 'mod1': '/sub_id/{visit_id}/mod1'}, ignore_exception=True)

What is the correct design so that the workflow is not failing when selectfile on mod2 does not match a file?
Currently I am creating a new workflow depending on what is in /{sub_id}/{visit_id}. Does not look ideal.

Optional question: Is there smart way of dealing with modo(t0) that would get co-register to himself in this worflow?

Thanks,

1 Like
2

@npann - we are hoping to address conditional logic in our next API

here are a few options we use at present:

  1. let branches fail and generate crashfiles.
  2. use a function node to guide things.
    • embed a workflow generator function inside a function node function
    • run the workflow or not depending on inputs.

Function Interfaces are quite powerful as they combine standard python control together with the ability to run workflows inside them.

3

Is there any update to this issue?

2 Likes