Hi Laura,

Yes, there is no need for any _per_run options when
there is only 1 run.

Sure 3 is a good number of PCs to remove, but nothing
is written in stone. Some studies have used larger
numbers (though I do not know about via afni_proc.py),
but from people I know who have done such things, 3
seems like a good number.

Yes, FreeSurfer would take much longer. But 3dDeconvolve
is not slowing down your processing, it is the non-linear
registration. Just adding -tlrc_NL_warp vastly increases
the processing time. And if your binaries support it,
the non-linear registration uses parallel processing to
speed it up already.

It is also possible that your computer is running out of
RAM during the registration step, which would greatly
slow it down. How much RAM do you have, and how large
are your datasets?

- rick