Hi Kara,

1) Yes, putting the align block before volreg is not so intuitive,
but is a nice way to go. The main goal is to not keep resampling
the EPI:

a. by running 3dvolreg
b. by aligning to the anat
c. by transforming to standard space (which you are not doing)

Since the latter 2 transformations are computed before 3dvolreg
is run, all transformations can be combined applied only once to
the EPI during the volreg block. It would certainly be possible
to put them in any order and still combine them, but doing it in
the volreg block puts the EPI transformations all in one place.


2) Yes, the align block does skull stripping. But that is only
used for registration (in your case, to the EPI and probably to
the SurfAnat dataset). Skull-stripping does not need to be so
good for these things. You should still (hopefully) see a good
alignment when reviewing with @ss_review_driver.


3) If alignment is failing, you will probably end up adding:

-align_opts_aea -cost lpc+ZZ -giant_move

to the afni_proc.py command. But you can more quickly test
via the following:

a. make a temporary test directory
b. make a copy of the align_epi_anat.py inputs, as stated in
your command
c. add the options: -cost lpc+ZZ -giant_move
d. run the command

If that works, start afni_proc.py over, and pass those options
via -align_opts_aea as stated above.

If that does not work, keep us posted.

- rick