Sparse matrix dimensions

[acihanckr]

Hello,

Thank you for providing this guide, it is very helpful. I am going through some parts of the code and I realized the following line in Case-study_Mouse-intestinal-epithelium_1906.ipynb raise an error:

adata.obs['mt_frac'] = adata.X[:, mt_gene_mask].sum(1)/adata.obs['n_counts']

When I investigated a little I realized adata.X is a 2d sparse matrix so it can't be divided by 1d series. Also, applying flatten or reshape directly doesn't work because it returns a numpy matrix which can't be flatten (or I couldn't). This might be a version issue, but my solution was replacing the line with

adata.obs['mt_frac'] = np.array(adata.X[:, mt_gene_mask].sum(1)).flatten()/adata.obs['n_counts']

Best

[LuckyMD]

Hi @acihanckr,

This notebook hasn't been updated in a while, so it could very well be that there are some issues due to dependency updates. We're actually working on a newer version of the tutorail in github.com/theislab/single-cell-best-practices