extracting spliced/unspliced matrix from annData object

[eynullazada]

Hi

I am running scvelo on my single cell RNA-seq data. I need to get the spliced and unspliced sparse matrix. I see it is saved as layers in and data object. When I export whole Anndata objcet to csv file ( adata.write_csvs('adata')), layers do not get extracted. I tried to specifically extract spliced and unspliced matrxes unsing other way like following, however it extracts the unreadable file:

>>> import pickle
>>> x=adata.layers['spliced']
>>> with open('spliced.txt', 'wb') as spliced:
...    pickle.dump(x, spliced)

I would like to export those highlighted data

>>> adata
AnnData object with n_obs × n_vars = 37001 × 54532
    var: 'Accession', 'Chromosome', 'End', 'Start', 'Strand'
    layers: 'ambiguous', 'matrix', **'spliced', 'unspliced'**

How can I do that?

Looking forward to hearing from you.

Khagani

[WeilerP]

@eynullazada, you could do the following

import pandas as pd

unspliced_df = pd.DataFrame(adata.layers['unspliced'], index=adata.obs_names, columns=adata.var_names)
unspliced_df.to_csv(PATH_TO_DATA / 'unspliced.csv')

spliced_df = pd.DataFrame(adata.layers['spliced'], index=adata.obs_names, columns=adata.var_names)
spliced_df.to_csv(PATH_TO_DATA / 'spliced.csv')

If your AnnData is loaded and PATH_TO_DATA is defined as a pathlib.Path pointing to the directory where your CSV files are to be saved, the code snipped should run as is.

[eynullazada]

Hi Weiler

Thanks so much for quick response. However, way you suggest doing it is not working and generating the following error:

>>> spliced_df = pd.DataFrame(adata.layers['spliced'], index=adata.obs_names, columns=adata.var_names)
Traceback (most recent call last):
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/managers.py", line 1678, in create_block_manager_from_blocks
    make_block(values=blocks[0], placement=slice(0, len(axes[0])))
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/blocks.py", line 3284, in make_block
    return klass(values, ndim=ndim, placement=placement)
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/blocks.py", line 2792, in __init__
    super().__init__(values, ndim=ndim, placement=placement)
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/blocks.py", line 128, in __init__
    "{mgr}".format(val=len(self.values), mgr=len(self.mgr_locs))
ValueError: Wrong number of items passed 1, placement implies 54532

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/frame.py", line 464, in __init__
    mgr = init_ndarray(data, index, columns, dtype=dtype, copy=copy)
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/construction.py", line 213, in init_ndarray
    return create_block_manager_from_blocks(block_values, [columns, index])
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/managers.py", line 1688, in create_block_manager_from_blocks
    construction_error(tot_items, blocks[0].shape[1:], axes, e)
  File "/mnt/ws/home/keynullazada/.conda/envs/p_env/lib/python3.7/site-packages/pandas/core/internals/managers.py", line 1719, in construction_error
    "Shape of passed values is {0}, indices imply {1}".format(passed, implied)
ValueError: Shape of passed values is (37001, 1), indices imply (37001, 54532)

Khagani

[eynullazada]

Unspliced and Spliced matrix is stored as Compressed Sparse Row format

[WeilerP]

Yes, I can see that. Didn't expect you to store a matrix of that size as CSV. You need to densify the matrix when defining the DataFrame

df = pd.DataFrame(adata.layers['spliced'].A, index=adata.obs_names, columns=adata.var_names)

[eynullazada]

great, thanks so much. That solved the issue.

I appreciate your help.