Incompatibility with new anndata 0.10.9

[mehrankr]

Issue

There is a new anndata release incompatible with scVI

Error message

> import scvi
ImportError: cannot import name 'AlignedViewMixin' from 'anndata._core.aligned_mapping' (/local_disk0/.ephemeral_nfs/envs/pythonEnv-41a17f4f-ae47-4ddc-a406-410dcbd282df/lib/python3.10/site-packages/anndata/_core/aligned_mapping.py)

Temporary fix

Pin down anndata:

pip install anndata==0.10.8

[canergen]

Hi, thanks for posting this. Please post the full error as it makes it easier to figure out what's causing this. I opened the issue for MuData as it seems like this is causing the issue: https://github.com/scverse/mudata/issues/77.

[canergen]

Fixed in MuData.

[jflusche]

Hello, I hope this message finds you well. I am getting the same error message having to do with scvi and anndata package incompatibilities when importing the newest version of both packages. I tried the temporary fix of downgrading anndata to version 0.10.8 but then got a message saying:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. scvi-tools 1.3.0 requires anndata>=0.11, but you have anndata 0.10.8 which is incompatible.

Any suggestions here? Thank you for taking the time to read this.

The entire original error message is pasted below for reference. `--------------------------------------------------------------------------- ImportError Traceback (most recent call last) Cell In[10], line 1 ----> 1 import scvi

File ~/.local/lib/python3.10/site-packages/scvi/init.py:11 8 from ._settings import settings 10 # this import needs to come after prior imports to prevent circular import ---> 11 from . import data, model, external, utils 13 from importlib.metadata import version 15 package_name = "scvi-tools"

File ~/.local/lib/python3.10/site-packages/scvi/data/init.py:4 1 from anndata import read_h5ad 2 from anndata.io import read_csv, read_loom, read_text ----> 4 from ._anntorchdataset import AnnTorchDataset 5 from ._datasets import ( 6 annotation_simulation, 7 brainlarge_dataset, (...) 23 synthetic_iid, 24 ) 25 from ._manager import AnnDataManager, AnnDataManagerValidationCheck

File ~/.local/lib/python3.10/site-packages/scvi/data/_anntorchdataset.py:20 17 import torch 19 from ._manager import AnnDataManager ---> 20 from ._utils import registry_key_to_default_dtype, scipy_to_torch_sparse 22 logger = logging.getLogger(name) 24 SparseDataset = (CSRDataset, CSCDataset)

File ~/.local/lib/python3.10/site-packages/scvi/data/_utils.py:15 13 from anndata.abc import CSCDataset, CSRDataset 14 from anndata.io import read_elem ---> 15 from mudata import MuData 16 from torch import as_tensor, sparse_csc_tensor, sparse_csr_tensor 18 from scvi import REGISTRY_KEYS, settings

File ~/.local/lib/python3.10/site-packages/mudata/init.py:3 1 """Multimodal omics analysis framework""" ----> 3 from ._core.mudata import MuData 4 from ._core import utils 5 from ._core.io import *

File ~/.local/lib/python3.10/site-packages/mudata/_core/mudata.py:21 19 from anndata import AnnData 20 from anndata.utils import convert_to_dict ---> 21 from anndata._core.aligned_mapping import ( 22 AxisArrays, 23 AlignedViewMixin, 24 AxisArraysBase, 25 PairwiseArrays, 26 PairwiseArraysView, 27 ) 28 from anndata._core.views import DataFrameView 30 from .file_backing import MuDataFileManager

ImportError: cannot import name 'AlignedViewMixin' from 'anndata._core.aligned_mapping' (/home/jupyter/.local/lib/python3.10/site-packages/anndata/_core/aligned_mapping.py)`

[ori-kron-wis]

please start a clean new conda env and install scvi-tools 1.3 , it will install the needed packages and should be ok

[jflusche]

Hello, thanks for the response! I tried this and after re-installing scvi (version 1.3.0) and anndata (version 0.11.3) I still am getting the same error. Do you have any other suggestions? Thanks.

[canergen]

Your issue here is MuData and not AnnData or scvi-tools. Can you verify that the issue arises when you just import MuData and then follow their installation guideline?

[jflusche]

The same issue does arise when I import MuData.

`import mudata as md

from mudata import MuData`

ImportError: cannot import name 'AlignedViewMixin' from 'anndata._core.aligned_mapping' (/home/jupyter/.local/lib/python3.10/site-packages/anndata/_core/aligned_mapping.py)

[jflusche]

Hi! Just following up on this issue. Do you have any suggestions on where to go from here? I get the same error when I follow MuData installation guideline. Should I reach out to the MuData team for assistance? Thank you for your help thus far.

ImportError: cannot import name 'AlignedViewMixin' from 'anndata._core.aligned_mapping'

[ori-kron-wis]

Like I suggested before, please start a clean new conda env and install scvi-tools 1.3 , it will install the needed packages for both anndata and mudata and should be ok. Use python 3.12.

[jflusche]

Hi, thank you for your suggestions! I wanted to follow up with what worked for solving this issue. I upgraded MuData to the newest version 0.3.1, which is compatible with scvi version 1.3.0. I also wanted to note that in order to load existing scvi models I had to downgrade pandas to version 1.5.3.

[joeyQX]

I have the same error, and there is no use of pip install anndata==0.10.8. Moreover, my MuData is the newest version 0.3.2

[ori-kron-wis]

@joeyQX Im not quite following, but anndata 0.10.x is at least 1 year old, and we can make certain it will be supported with most recent scvi-tools (e.g 1.4.x). I suggest starting a clean environment and setting up scvi-tools, which will automatically install all other needed packages correct versions. Let me know if you have other questions.

[joeyQX]

@joeyQX Im not quite following, but anndata 0.10.x is at least 1 year old, and we can make certain it will be supported with most recent scvi-tools (e.g 1.4.x). I suggest starting a clean environment and setting up scvi-tools, which will automatically install all other needed packages correct versions. Let me know if you have other questions.

thx!