Open lf96abc opened 3 months ago
adata.obsm['fragment_paired'] stores single base resolution count matrix in a compact format as described here: https://kzhang.org/SnapATAC2/api/_autosummary/snapatac2.pp.import_data.html#snapatac2.pp.import_data.
It is not easy to extract this matrix as it is huge. Instead, you can extract a low resolution version of it after calling pp.add_tile_matrix
. If you do want to get a single base resolution count matrix, you can call pp.add_tile_matrix
with bin_size=1
. But this will likely run into out-of-memory.
adata.obsm['fragment_paired'] stores single base resolution count matrix in a compact format as described here: https://kzhang.org/SnapATAC2/api/_autosummary/snapatac2.pp.import_data.html#snapatac2.pp.import_data.
It is not easy to extract this matrix as it is huge. Instead, you can extract a low resolution version of it after calling
pp.add_tile_matrix
. If you do want to get a single base resolution count matrix, you can callpp.add_tile_matrix
withbin_size=1
. But this will likely run into out-of-memory.
so is there any other way to get around this so that fragment generated and annotation done using snapatac2 can be used for pycisTopic?
What information exactly do you need?
What information exactly do you need?
The starting input required in th pycisTopic tutorial are the fragment files and annotations from each cell. Since, both are already stored in the snapatac anndata, I was wondering if the fragments and cell_data could be extracted and used as an input for pycisTopic.
Hello,
Thanks for the great package!
I am trying to extract the fragment count matrix to implement with the following notebook:
https://github.com/aertslab/pycisTopic/blob/old/notebooks/Toy_melanoma-RTD.ipynb
However, I cannot see where to extract the fragment count matrix. I believe this is stored in adata.obsm['fragment_paired'], but I cannot see how to extract the fragment names.
Thank you for your help