how to get mapping matrix from ST alignment task

[yuling999666]

Hi @ilibarra , I would like to ask how to get mapping matrix for alignment? Do you compute it? I didn't find it in your code. Thanks.

[yuling999666]

Hi @MUCDK , I have solved this issue: solutions = ap.solutions

Access the computed solutions

for cur_idx in range(1,len(slice_ids)):
index=str(slice_ids[cur_idx-1])+'_'+str(slice_ids[cur_idx])
map_path=osp.join('/home/data3/yuling/xenium_region2_5599/moscot/mHy', index+'_AlignmentPi.pkl')
with open(map_path, "wb") as f:
pkl.dump(solutions[tuple([str(cur_idx-1), str(cur_idx)])].transport_matrix, f)
But right now, I have another question. After I read about method for the paper, I think alignment method is a deterministic algorithm, no randomness will be generated during this course. Is that correct? Thank you!

[MUCDK]

Hi @yuling999666 ,

This is correct!

And yes, the mapping is determinisstic.

[yuling999666]

Hi @yuling999666 ,

This is correct!

And yes, the mapping is determinisstic.

Hi @MUCDK , thank you for your reply. Do you mean the whole process for alignment including getting aligned spatial coordinates is deterministic? Thanks.

[MUCDK]

Yes, note that (F)GW (the underlying alg.) is not guaranteed to converge to the global minimum and the solution is not unique, but it's still deterministic. Do you observe non-deterministic behaviour or why are you asking?