Issue working with data as tuples

[enicolaisen]

Description of issue

Hi!

Thank you very much for making this tool available!

I would like to let you know that I have faced an issue when working with data stored in tuples (tuples of arrays). According to the documentation, functions such as neuromaps.nulls.alexander_bloch can deal with data in format tuple. However, I get the following error with the following code:

import numpy as np
x = np.arange(10242)
x.shape #(10242,)

rotated = nulls.alexander_bloch((x,x), atlas='fsaverage', density='10k',n_perm=3, seed=0)

Error:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/data/project/pet_neurochemistry/enic/petparti/code/pet_parti/lib/python3.10/site-packages/neuromaps/nulls/nulls.py", line 145, in alexander_bloch
    return load_data(data)[spins]
IndexError: index 18787 is out of bounds for axis 0 with size 10242

We were able to track the issue back to the function load_data, which does not return the data in the expected format:

y = load_data((x, x))
>>> y.shape #(10242, 2)

We were able to find a work-around by doing this:

img_nii=np.hstack(img_nii)

But anyway we thought it would be nice to let you know.

Thanks again for the tool!

Best,
Eli

Code of Conduct

• I agree to follow the neuromaps Code of Conduct

[VinceBaz]

Hi @enicolaisen

Thank you for letting us know!

We apologize for the confusion. Currently, when no parcellation is provided, the function accepts surface-based images or path-like objects for these images and tuples of them, but it does not accept array-like data structures (e.g. numpy arrays as in your example), nor does it accept tuples of array-like structures.

You however show that one line of code is apparently all we need to fix this, so allowing users to input array-like structures is definitely something that we should consider. We'll try to implement it and keep you posted! 😁

Best,
Vince