Closed akhst7 closed 2 years ago
Hey akhst7!
I'll take a look at it asap. In the meantime, could you tell me which version of R, Python and anndata (the python kind) you're using?
Robrecht
@rcannood , my R, Python, anndata python are as follows;
> sessionInfo()
R version 4.1.2 (2021-11-01)
Platform: aarch64-apple-darwin20 (64-bit)
Running under: macOS Monterey 12.2.1
Matrix products: default
LAPACK: /Library/Frameworks/R.framework/Versions/4.1-arm64/Resources/lib/libRlapack.dylib
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] anndata_0.7.5.3 reticulate_1.24 SeuratObject_4.0.4
loaded via a namespace (and not attached):
[1] compiler_4.1.2 here_1.0.1 assertthat_0.2.1 R6_2.5.1 Matrix_1.4-0 rprojroot_2.0.2 cli_3.2.0
[8] tools_4.1.2 Rcpp_1.0.8 grid_4.1.2 jsonlite_1.8.0 rlang_1.0.2 png_0.1-7 lattice_0.20-45
> py_config()
python: /opt/homebrew/Caskroom/miniforge/base/envs/velocity/bin/python
libpython: /opt/homebrew/Caskroom/miniforge/base/envs/velocity/lib/libpython3.10.dylib
pythonhome: /opt/homebrew/Caskroom/miniforge/base/envs/velocity:/opt/homebrew/Caskroom/miniforge/base/envs/velocity
version: 3.10.2 | packaged by conda-forge | (main, Feb 1 2022, 19:29:01) [Clang 11.1.0 ]
numpy: /opt/homebrew/Caskroom/miniforge/base/envs/velocity/lib/python3.10/site-packages/numpy
numpy_version: 1.21.5
❯ conda list 08:49:48 PM
# packages in environment at /opt/homebrew/Caskroom/miniforge/base/envs/velocity:
#
# Name Version Build Channel
anndata 0.7.8 pypi_0 pypi
I really appreciate your help !!
I see -- the problem lies with scipy deprecating classes like scipy.sparse.csc.csc_matrix
and renaming them to scipy.sparse._csc.csc_matrix
. I feel like this is something that should be solved in the reticulate conversion functions, but I can implement a workaround in anndata for now if need be.
(If you need something that works now, a temporary workaround could also be downgrading to scipy <1.8.0)
I created a PR at rstudio/reticulate#1173.
@akhst7 if you want, you could try installing my fork of reticulate to see if this solves your issue: r remotes::install("rcannood/reticulate")
.
@rcannood, thanks for digging this out. So both downgrading scipy (1.7.3) and installing your reticulate commit (with scipy 1.8.0) works. BTW, how did you figure out the culprit was scipy ?
@rcannood Hi! I'm facing the "Error in dimnames(out) <- dimnames(self) : 'dimnames' applied to non-array" issue right now and trying to solve it by "remotes::install_github("rcannood/rstudio")", but don't have permissions to access the repo. I would like to ask for permission to install your reticulate. Thanks!
@mayingxiang Hi Andy. Oops, I meant to say remotes::install_github("rcannood/reticulate")
. I've edited my post in the meantime.
However, the PR has been merged into rstudio/reticulate
, so you could also simply install the development version of reticulate: remotes::install_github("rstudio/reticulate")
.
@rcannood Thank you! It's working now.
This is probably related to the issue-803759711. I imported h5ad just as I normally do;
ann<-read_h5ad("adata.h5ad")
and when I checked X as follows;
I guess all the space matrix including the ones in layers (e.g. spliced) is not accessible form R. It was working fine a month ago. It may have some to do with recent updates of h5dy (3.6.0) ?