Performing seurat-style data integration on data analysed using scanpy?

[ydang]

I have done an analysis using scanpy and related sc-verse pipelines of a large number of separate data sets (8). These represent three different time points and for each time point I have two conditions. I would like to integrate this data, and personally found the seurat integration pipeline to be best for doing this. However, as far as I know there is no python equivalent of this anchor-based integration approach, am I right? Do you have other suggestions for how to integrate the data using a python-based method?

[ivirshup]

I’d look into scanorama. It integrates datasets in a similar way, and compares favorably to Seurat in benchmarks.

[Echoo-ranger]

Hi, I have a very similar question. I am handling a number of cell ranger output files from different experiments and integrating them to together.

I think Integration is just reading all the files in a for loop and concatenating them. But the issue I am having is adding metadata of the experiments? like type of organ, etc? Is there a way to add those

[ivirshup]

I think this question is a bit different, since it’s not about accounting for batch effects in the data. I’ve addressed this in your other post here:

How to add additional metadata on multiple different single cell RNA files from different experiments, such as organ, sample number etc

Hey @Echoo-ranger, Could you add this information as columns of the obs dataframe before concatenation? If there is only one condition per cell ranger output, you could also add this information during concatenation with the label argument. Then the code should probably look something like: import scanpy as sc import anndata as ad # Read in each experiment experiment_dirs = ["organ1", "organ2", ...] experiment_adatas = { expr_name: sc.read_10x_h5(f"path/to/{expr_name}/outs/filtered_featur…