Connect a MapNode to a workflow in parallel

1

Hello:
I have a question for nipype parallelization, for example, I have a Pipeline to run diffusion preprocessing, but the pipeline is written in sequence which means one time for one subject, So if I wanna parallelize it with 10 subjects, can I connect a MapNode with the workflow?

More specifically:

Node to distribute the subjects in parallel

distribute_node = npe.Node(nutil.IdentityInterface(name="Distributesubjects",
                                interface=nutil.IdentityInterface(
                                    fields=['in_dwi', 'in_bval', 'in_bvec', 'in_fmap_magnitude', 'in_fmap_phasediff'],
                                    mandatory_inputs=True)),
                       name='b0_dwi_split')

    full_pipe = diffusion_preprocessing_phasediff_fieldmap()

It seems that this does not work, do you have this functionalities?

Thanks in advance

2

Iterables seems like a better fit for such use case (parallelizing over participants) - see http://nipype.readthedocs.io/en/latest/users/mapnode_and_iterables.html

3

Thanks, I made it with iterables, thanks anyway