If you are already familiar with some of the AFNI processing steps, you might want to use @auto_tlrc to put an anatomical, B0 or FA dataset in Talairach or MNI space and then use align_epi_anat.py to align the DWI or FA data to the anatomical and apply the combined transformation to put either in the standard space:

@auto_tlrc anat+orig
align_epi_anat.py -anat anat+orig -epi FA+orig -cost lpa -epi_base 0 _suffix _al2anat -tlrc_apar anat+tlrc -prep_off

or

# to do motion correction of DWI data at the same time
align_epi_anat.py -anat anat+orig -epi DWI+orig -cost lpa -epi_base 0 _suffix _al2anat -tlrc_apar anat+tlrc -volreg_method 3dAllineate

Note transforming the DWI data means you will have also have to rotate the gradient vectors before computing the tensor. See previous threads on how to do this:

[afni.nimh.nih.gov]
<[afni.nimh.nih.gov];

For you second question, use 3dUndump -srad to create spheres based on the anat+tlrc dataset, and then put back into original space using one of the methods suggested here:

[afni.nimh.nih.gov]
<[afni.nimh.nih.gov];


You can also check approaches like this one that combine AFNI with DTI-Query:
[openwetware.org]
[openwetware.org]

or the Tortoise package:
[www.tortoisedti.org]
[www.tortoisedti.org]