Conditional run of nodes

[Michael]

Hi,

How would be the most efficient and elegant way to run a node depending on some logical conditions? According to logical tests, different outcomes could be:
- skipping/running a node
- re-running the same node (with different parameters or not)
- running a selected node among a group with similar functions (and inputs/outputs)
- skipping/running a different workflow before coming back into current workflow

[Satrajit Ghosh]

hi michael,

these are all excellent questions to which we don't quite have a simple answer (especially doing this at runtime).

- skipping/running a node 

- running a selected node among a group with similar functions (and inputs/outputs)

- skipping/running a different workflow before coming back into current workflow

we have generally taken care of these scenarios through function nodes that we insert into he pipeline. for example here is one that compares different realignment algorithms. here is an example:

https://github.com/INCF/BrainImagingPipelines/blob/master/bips/workflows/gablab/wips/scripts/modular_nodes.py#L245

https://github.com/INCF/BrainImagingPipelines/blob/master/bips/workflows/gablab/wips/scripts/base.py#L223

 

- re-running the same node (with different parameters or not)

new_anode = anode.clone(name="cloned_anode")

 

any thoughts, pull requests welcome.

cheers,

satra

[Michael]

Thank you for these example of function nodes, they will be very useful.