[PyMOL] Aligning a object to camera axis

[Shyam S. <sa...@ca...>]

Hello --

I am trying to transform an object to align it's major axis with the x-axis
at z = 0 in the camera coordinate space. Sample pdb file (just a simple
scale bar) is attached.

My thought process is as follows:

1. Convert <1, 0, 0> in camera space to model space (vector A)
2. Find vector that corresponds to object's major axis (vector B)
3. Find transformation matrix C that rotates to vector B to vector A
4. Use transform_selection with C


Unfortunately, my attempts are without success (see below), perhaps due to
my lack of understanding of the pymol coordinate spaces.

I see saw two potentially relevant threads here, but I wasn't able to pull
things together successfully:

https://www.mail-archive.com/pym...@li.../msg14278.html
https://sourceforge.net/p/pymol/mailman/message/10098601/


If you have any thoughts here, please let me know!

Thanks,
Shyam


M_to_model = np.array(cmd.get_view()[:9]).reshape(3,3)
def to_model(X): return np.matmul(X, M_to_model)

# 1. xaxis in model coordinates
xaxis_mod = to_model([10, 0, 0])

# 2. major axis of scale in model coordinates
scale_orient = cmd.get_extent("scale_25")
scale_orient = np.diff(scale_orient, axis=0)[0]
print(xaxis_mod, scale_orient)

# 3. matrix to rotate scale to axis
# From scipy.spatial.transform.rotation
#
https://github.com/scipy/scipy/blob/01d8bfb6f239df4ce70c799b9b485b53733c9911/scipy/spatial/transform/rotation.py#L1854-L1982
rot, _ = align_vectors([scale_orient], [xaxis_mod])

# 4. transform scale
def reformat_3x3(rot): \
return [*rot[0], 0, \
*rot[1], 0, \
*rot[2], 0, \
0, 0, 0, 1]
cmd.transform_selection("scale_25", reformat_3x3(rot), homogenous=1)