This should be pretty easy: look at how randomise is called in the dual_regression script. It has to be run on each of the stage2 output files you're interested in (there's one for each ICA component).

I'd also recommend first checking how long the first stages took: it may be that the randomise calls took much of the processing time. I'd suggest assessing your dual regression call (e.g. n_perm) and/or finding a faster computer: this seems very slow for a regular sized dataset.

On 10 May 2014 15:24, Patrick Pflanz <[log in to unmask]> wrote:

Dear FSL experts,

I have been running the dual regression command at one of our local computers in psychiatry since easter. However, today there has been a computer crash, so the analysis has been interrupted. Basically, all stages of the dual regression have already run apart from the last step, which includes the randomize command. If I have to start the analysis from scratch, it will take quite a long time to compute everything from square one. I was therefore wondering, if there is a chance to resume the analysis, and to restart at the last component before the computer crashed. This would really be helful, because it would save quite a lot of time.

Many thanks for any help you are able to offer.

With best regards,
Patrick Pflanz