Export to Anndata (after integration)

[DaminK]

Thanks for all the work you put in this cool package!

I was wondering if you are considering adding an Anndata export / interface to easily transfer aligned scRNA datasets back into a single anndata object for further downstream analysis. If this is available somewhere please point it out as I couldn't find it in the source code.

It should be feasible and would enable many less technical people to use your package for preprocessing for other use-cases.

A simple (inefficient) approach to implement this is as follows:

#Integration with LIGER
adatas_liger = pyliger.create_liger([adata1,adata2,adata3, ...])
... normalize, select_genes, scale_not_center, optimize_ALS, quantile_norm ...

#Export in a shared Anndata & store transformed gene expression
adata_shared = ad.concat([adata1,adata2,adata3, ...],axis =0, join="outer") 
adata_shared.obsm["X_liger"] = np.zeros(adata_shared.X.shape)
for i,gene in tqdm(enumerate(adata_shared.var.index)):
    adata_shared.obsm["X_liger"][:,i] =  pyliger.plotting._utilities.get_gene_values(adatas_liger, gene)

I have a small working prototype for my own export, if you are interested in merging this (after optimizing it by not using calling get_gene_values for indivdual genes and rather perform this operation parallelized on all genes), let me know and I will open a pull request. One might also want to include the computed factorizations (as obsm and varm according to best practizes for other factorizations in anndata)