QUESTION: Subset expression matrix from custom genes

[cartal]

Hi,

I have asked this question before in Scanpy, but I wasn't sure I made it clear. So I'm giving it a try again:

Say I have the PBMC 3K dataset, and after clustering and DEG in Scanpy, I have 120 genes specific for cluster 1 and 80 genes specific for cluster 3. Now, what I would like is to take these 200 genes and export them as an expression matrix (csv or tab), where rownames are the gene symbols (i.e: IFNG, IL10, etc) and colnames are the cell ids (barcodes) present within each cluster.

How do I get this from an h5ad file?

Thanks!

[falexwolf]

Why isn't this answer (scverse/scanpy#262 (comment)) working for you?

Obviously, you'd need to subset the anndata object before:

import scanpy.api as sc
adata = sc.datasets.pbmc68k_reduced()
# this writes everything, the data matrix and annotations end up in separate files
adata[:, ['HES4', 'TNFRSF4', 'SSU72']].copy().write_csvs('./test/', skip_data=False)
# this writes only the data matrix
adata[:, ['HES4', 'TNFRSF4', 'SSU72']].to_df().to_csv('./matrix.csv')

Hope this helps!

PS: Please, next time reference your previous issue. Also, provide a small example such as the one above.