Error in inline usage about solo HOT 7 CLOSED

Hi Jens,

Let me take a look. It will probably just convert the sparse matrix under the hood into a dense matrix because I don't think cell hashing matrices should cause memory issues. Its interesting the

hdata.X = hdata.X.todense()

didnt work.

from solo.

Funnily, saving the hdata object with force_dense = True (as_dense = True in scanpy >=1.5.1) and reloading the object seems to do the trick.

Nevertheless, can hashsolo be easily written to take sparse matrices as well?

from solo.

Any update? Maybe a workaround? :-)

from solo.

@jenzopr I merged a fix just now let me know if it works.

Some of the indexing I do in hashsolo doesn't work with a dense matrix so the code needs to be:

if issparse(cell_hashing_adata.X):
    cell_hashing_adata.X = np.array(cell_hashing_adata.X.todense())

from solo.

If you need this fix released to pypi let me know otherwise you can install the fix from cloning the master branch and pip installing from there

from solo.

Thats great! Thanks a lot for your support 🎉

from solo.

Same error still appears to happen in the newest version. Saving the .h5ad file using as_dense='X' option, followed by reading it back in again seems to solve the issue, while

adata.X = adata.X.todense()

does not work.

from solo.