Is there a nipype wrapper for randomise_parallel? I see the randomise interface, but when I tried to run the code, it seems to be running serially (i.e. randomize not randomise_parallel). If not, is there a similar functionality provided in one of the other packages for this sort of functionality?


It seems to potentially have existed somewhere in the interface… and I see an old test in the interfaces/fsl/tests/test_dti.py file in the latest 0.12 Build…

Also google led me to this, but this is from an old build of nipype…

Just wanted to bump this again… Was there something fundamentally weird about Randomise_Parallel that had it no longer be supported by NiPype?

@dagutman - right now we don’t have it.

  • in the fsl world PALM has essentially replaced randomise, and we will likely have a formal interface for PALM shortly.
  • regarding the parallel aspect of randomise, it was a shell script that used SGE to break up the pieces, and we never reimplemented that process. at present we are likely to go with PALM and optimize for that and the updated afni tool for permutation testing.

we have a nipype workshop in 9 days and can propose the randomise parallel as an example workflow construction project.

Thanks Satra— I’ll play around more with PALM, I just had some old TBSS Scripts I was trying to port over to NiPype for demonstration purposes…

I am sending a student to attend the conference so I’ll make sure he brings that up…

Good day, sorry but i’m having some trouble using randomise, allow me to ask this, i’m doing 14 contrasts for 5 EV’s trying to do a covariance analysis, and when run the randomize script and enter 10,000 permutations , the t-tests 9 and 10 do 1764 permutations, why is this hapening?

Thanks for your time