type ERROR

[nicetheworld]

Hi Thanks for this great resource，
when I use map functation of hnoca tools, I sucessfully run

 early_stopping_kwargs = {
    "early_stopping_metric": "val_loss",
    "mode": "min",
    "threshold": 0,
    "patience": 10,
    "reduce_lr": True,
    "lr_patience": 5,
    "lr_factor": 0.1,
}

mapper.map_query(
    adata_query_v1_hvg,
    retrain="partial",
    max_epochs=20,
    batch_size=256,
    unlabeled_prototype_training=False,
    n_epochs=10,
    pretraining_epochs=9,
    early_stopping_kwargs=early_stopping_kwargs,
    eta=10,
    alpha_epoch_anneal=10,
)

but when it comes to mapper.compute_wknn(k=100)
it get an error :

AttributeError Traceback (most recent call last)
Cell In[44], line 1
----> 1 mapper.compute_wknn(k=100)

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/hnoca/map/mapper.py:198, in AtlasMapper.compute_wknn(self, ref_adata, k, query2ref, ref2query, ref_rep_key, query_rep_key, weighting_scheme, top_n)
196 self.ref_adata = ref_adata if ref_adata is not None else self.ref_adata
197 ref_latent = self._get_latent(self.query_model, self.ref_adata, key=ref_rep_key)
--> 198 query_latent = self._get_latent(
199 self.query_model, self.query_adata, key=query_rep_key
200 )
202 wknn = get_wknn(
203 ref=ref_latent,
204 query=query_latent,
(...)
209 top_n=top_n,
210 )
212 self.wknn = wknn

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/hnoca/map/mapper.py:152, in AtlasMapper._get_latent(self, model, adata, key, **kwargs)
150 return model.get_latent_representation(adata, **kwargs)
151 if self.model_type == "scpoli":
--> 152 return model.get_latent(adata, **kwargs)

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/scarches/models/scpoli/scpoli_model.py:354, in scPoli.get_latent(self, adata, mean)
352 subsampled_indices = indices.split(512)
353 for batch in subsampled_indices:
--> 354 x_batch = x[batch, :]
355 if sparse.issparse(x_batch):
356 x_batch = x_batch.toarray()

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/scipy/sparse/_index.py:30, in IndexMixin.getitem(self, key)
29 def getitem(self, key):
---> 30 index, new_shape = self._validate_indices(key)
32 # 1D array
33 if len(index) == 1:

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/scipy/sparse/_index.py:221, in IndexMixin._validate_indices(self, key)
219 index_1st.append(idx)
220 prelim_ndim += 1
--> 221 elif (ix := _compatible_boolean_index(idx, self.ndim)) is not None:
222 index_1st.append(ix)
223 prelim_ndim += ix.ndim

File /ssd/software/miniconda3/envs/hnoca/lib/python3.10/site-packages/scipy/sparse/_index.py:428, in _compatible_boolean_index(idx, desired_ndim)
425 # since first is boolean, construct array and check all elements
426 idx = np.asanyarray(idx)
--> 428 if idx.dtype.kind == 'b':
429 return idx
430 return None

AttributeError: 'torch.dtype' object has no attribute 'kind'

[Marius1311]

Hi @nicetheworld, can you please update to v0.2.1 of HNOCA-tools, and check whether the error persists? If it does, please also report relevant package versions in the issue. Thanks!