Mixscape data type error

[AlejandroTL]

Report

Hello!

When I'm trying to reproduce the tutorial with a different dataset, I receive the following error in the function Mixscape.mixscape.

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[70], line 1
----> 1 mixscape_identifier.mixscape(adata = adata, control = 'control', labels='perturbation', layer='X_pert')

File ~/miniconda3/envs/pertpy/lib/python3.9/site-packages/pertpy/tools/_mixscape.py:244, in Mixscape.mixscape(self, adata, labels, control, new_class_name, min_de_genes, layer, logfc_threshold, iter_num, split_by, pval_cutoff, perturbation_type, copy)
    240 vec = np.mean(X[guide_cells][:, de_genes_indices], axis=0) - np.mean(
    241     X[nt_cells][:, de_genes_indices], axis=0
    242 )
    243 # project cells onto the perturbation vector
--> 244 pvec = np.sum(np.multiply(dat.toarray(), vec), axis=1) / np.sum(np.multiply(vec, vec))
    245 pvec = pd.Series(np.asarray(pvec).flatten(), index=list(all_cells.index[all_cells]))
    246 if n_iter == 0:

AttributeError: 'numpy.ndarray' object has no attribute 'toarray'

Nevertheless, it doesn't happen when running with the dataset of the tutorial. I think that the problem is that Mixscape.pert_sign does not create .layers['X_pert'] as a sparse matrix, but as a normal numpy array.

Thanks!

Version information

No response

[xinyuejohn]

@AlejandroTL Thanks for finding out this bug! I will solve this as soon as possible.