Daniel may not get to this today, so I will chime in.

To back up one step, let me note that there are 3 aspects to
this that vary, and often none of them are fully reported.
You may already understand this all, but it is good to post
it in any case.

Background:

a. the dataset space, which is essentially defined by the template

When one says the data is aligned to a particular template (not
an atlas, but an anatomical registration template), then that
template essentially becomes the definition of the space. This
implies, for example, that there are multiple MNI and TLRC spaces.
In a general sense, they are typically viewed as being the same,
but it is useful to report the actual template used (which many
people currently do).


b. the data orientation, such as RAI

Fortunately we do not yet include oblique data here.

Ignoring that, there are 48 possibilities starting from one of
the 6 right/left or A/P or I/S directions for the first data
direction, then starting from one of the remaining 4 for the
next, then one of the remaining 2 for the final direction.

Or, from the direction that Nick stated, think of it as one of
3 axes to start along, times one of 2 next and 1 final, all
times 2^3 choices for which direction to go along each axis.


c. how the orientation is actually stated

When we say "RAI", we mean right to left, A to P and I to S.
RAI is short-hand for RL,AP,IS. So the "R" is our starting
location, or direction source. The R,A,I sides are the
negative sides of the axes.

But some people call that "LPS", because they state the
orientation based on the positive directions.

======================================================================
On to your questions...

1. The CA_N27_ML dataset is TT_caez_ml_18+tlrc, which is
in TLRC space (implied by the leading "TT").

Note that this dataset should not be used in an @auto_tlrc
command, as it is just a set of macro labels, and not an
anatomical image (e.g. a T1 map).

In any case, it is not in MNI space.

Consider instead using the MNI_caez_N27+tlrc template.

2. The afni coordinates are based on the template, point (a)
above. Those numbers mean nothing if the space is not known
(such as the case of +orig space, where there is no reference).
In your case, they should be in TLRC space.

Also, afni defaults to showing RAI orientation, though you
can change that to LPI by right-clicking on RAI=DICOM or via
env vars.

3. Most AFNI programs now state coordinates in LPI orientation,
since that is most common in the literature. Unfortunately that
makes our software a little inconsistent. The 3dCM output is
also in LPI orientation, so -13 -64 -4 would mean 13L 64P 4I
(to correspond with your 13 64 -4 RAI coords).

4. If your template were in MNI space, then those coords would
be in MNI space. But I think they are in TLRC space. See (1).

5. The jump-to-MNI in the afni GUI makes the assumption that
it needs to transform from MNI to TLRC. But if you are already
in MNI space, then you should not alter coords or use that
interface. That is a little confusing, I know.

Basically, if your data is already in MNI space, then any
jump-to should take you to those same coordinates. The real
question remains, what template was the data aligned to?

6. Hopefully this is clear now. If you are already in MNI
space (which you are not, it seems), do nothing (but have afni
state LPI order).

Note that the "whereami" output will try to do the conversions
for you, based on the space it thinks the data is in.

- rick