Errors encountered when running the function sciPENN_API(...)

[CCCC1800]

Hi, I followed your tutorial "tutorial.ipynb" at your website https://drive.google.com/drive/folders/1iY4s76UYNMFvF6v3XN4JxD9gM77NIxoH. However, I got an error when running the following function:

sciPENN = sciPENN_API(gene_trainsets = [adata_gene_train], protein_trainsets = [adata_protein_train],
gene_test = adata_gene_test, train_batchkeys = ['donor'], test_batchkey = 'donor',
type_key = 'celltype.l3')

I got the following error:

---

TypeError Traceback (most recent call last)
TypeError: expected dtype object, got 'numpy.dtype[float64]'

The above exception was the direct cause of the following exception:

SystemError Traceback (most recent call last)
/var/folders/vn/xxwlw_f508gclgvh8ldr47sw0000gn/T/ipykernel_72171/2889813855.py in
1 sciPENN = sciPENN_API(gene_trainsets = [adata_gene_train], protein_trainsets = [adata_protein_train],
2 gene_test = adata_gene_test, train_batchkeys = ['donor'], test_batchkey = 'donor',
----> 3 type_key = 'celltype.l3')

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/sciPENN/sciPENN_API.py in init(self, gene_trainsets, protein_trainsets, gene_test, gene_list, select_hvg, train_batchkeys, test_batchkey, type_key, cell_normalize, log_normalize, gene_normalize, min_cells, min_genes, batch_size, val_split, use_gpu)
35 gene_list, select_hvg, cell_normalize, log_normalize, gene_normalize, min_cells, min_genes)
36
---> 37 genes, proteins, genes_test, bools, train_keys, categories = preprocess(*preprocess_args)
38
39 self.proteins = proteins

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/sciPENN/Preprocessing.py in preprocess(gene_trainsets, protein_trainsets, gene_test, train_batchkeys, test_batchkey, type_key, gene_list, select_hvg, cell_normalize, log_normalize, gene_normalize, min_cells, min_genes)
145
146 sc.pp.highly_variable_genes(tmp, min_mean = 0.0125, max_mean = 3, min_disp = 0.5,
--> 147 n_bins = 20, subset = False, batch_key = 'batch', n_top_genes = 1000)
148 hvgs = tmp.var.index[tmp.var['highly_variable']].copy()
149 tmp = None

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/scanpy/preprocessing/_highly_variable_genes.py in highly_variable_genes(adata, layer, n_top_genes, min_disp, max_disp, min_mean, max_mean, span, n_bins, flavor, subset, inplace, batch_key, check_values)
472 n_top_genes=n_top_genes,
473 n_bins=n_bins,
--> 474 flavor=flavor,
475 )
476

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/scanpy/preprocessing/_highly_variable_genes.py in _highly_variable_genes_single_batch(adata, layer, min_disp, max_disp, min_mean, max_mean, n_top_genes, n_bins, flavor)
200 X = np.expm1(X)
201
--> 202 mean, var = materialize_as_ndarray(_get_mean_var(X))
203 # now actually compute the dispersion
204 mean[mean == 0] = 1e-12 # set entries equal to zero to small value

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/scanpy/preprocessing/_utils.py in _get_mean_var(X, axis)
6 def _get_mean_var(X, *, axis=0):
7 if sparse.issparse(X):
----> 8 mean, var = sparse_mean_variance_axis(X, axis=axis)
9 else:
10 mean = np.mean(X, axis=axis, dtype=np.float64)

~/opt/anaconda3/envs/scipennenv/lib/python3.7/site-packages/scanpy/preprocessing/_utils.py in sparse_mean_variance_axis(mtx, axis)
40 )
41 else:
---> 42 return sparse_mean_var_minor_axis(mtx.data, mtx.indices, *shape, np.float64)
43
44

SystemError: CPUDispatcher(<function sparse_mean_var_minor_axis at 0x7fa7200328c8>) returned a result with an error set.

[CCCC1800]

Before encountering this error, I already had all the dependencies installed the same version as the software requirement written on your github.
My python version is 3.7.0.
In my conda environment "scipennenv", I had the following package versions:
torch=1.13.1
scanpy=1.9.3
pandas=1.1.5
numpy=1.21.6
scipy=1.7.3
tqdm=4.66.1
anndata=0.8.0
numba=0.50.0

Also, the active kernel I use is:
Python [conda env:scipennenv]

The reported error is encountered based on these settings.

[fisherj-2212]

I am also experiencing this error using a conda env built from the .yaml provided.

[fisherj-2212]

For future reference, I have gotten the software working by doing the following. It appears updating numba to a higher version than indicated in the dependencies is required. To do this I created a conda env from this .yml file which simply sets up the env with a suitable version of python and pip

name: scipenn_env_testing
channels:
  - conda-forge
  - defaults
dependencies:
  - python=3.7.0
  - pip>=21.2.4
prefix:  /export/home/jfisher2/mambaforge/envs/scipenn_env_testing

and then ran the following package installation commands in terminal

  pip install llvmlite==0.36 --ignore-installed
  pip install sciPENN --ignore-installed
  mamba install numba==0.53.0

Not the most elegant solution, but the sciPENN_API function does now run as expected.

[Genshin-Impact-king]

Is there any other solution?