theislab/scvelo

scv.read(adata.h5ad) does not load spliced and unspliced layers

mdu4003 opened this issue · 3 comments

Hello,

I am trying to do scvelo analysis on snRNAseq. I prepared the adata.h5ad using kb-python with a index built with the --workflow nucleus. I used that workflow also in kb count. The log shows that both matrixes are generated and combined in the adata.h5ad file. But when I read it with scv.read, the object shows 0 Layers.
...
This is the log for the counting:
TCD_sn_KB.out.8877564.txt

This is the command in scvelo:
adata_Epi_TCD35_P = scv.read("/Users/diazmeco/Desktop/SCU/TCD35kb_P/counts_unfiltered/adata.h5ad", chunk_size=100000)
adata_Epi_TCD35_
AnnData object with n_obs × n_vars = 182745 × 33696
obs: 'n_genes'

This is my scvelo:
scvelo.txt

Thank you very much!
Best regards,
Angeles

I started from scratch with new version of kb-python and workflow=nac.
The only issue now is that I have nascent (unspliced) and mature (spliced) which are not recognized by scv.pp.filter_and_normalize.
scv.pp.filter_and_normalize(adata_Epi_TCD35_P_filt, min_shared_counts=20, n_top_genes=2000)
WARNING: Could not find spliced / unspliced counts.

Do you have any solution? Should we re-name the layers somehow? Thanks!

I copied the layers with a new name:
adata_filt.layers["unspliced"] = adata_filt.layers["nascent"]
adata_filt.layers["spliced"] = adata_filt.layers["mature"]
It seems to be solved.
Can you add a modification for scvelo to recognize this new nomenclature?

Won't be implemented for now.