Understanding of tutorial-predict

[raphaelmourad]

Dear,

First of all I want to thank your effort for making notebooks and an easy to install python package. I can say everything has been very easy to run properly.

I am trying to use your notebook tutorial-predict.

Here, it is written use "all_context_adata" as context but from what the paper and what you wrote in the notebook, I would use instead "ctrl_adata" which are T cells as context for stack-generation. As expected "test_adata" are T cells with perturbed and control conditions.

Indeed you wrote : "The context AnnData consists of T cells under each drug-perturbation condition, while the test AnnData contains control myeloid and B cells."

Did i misunderstand something?

Regards
Raf

[andrewdchen]

Hello!

This is awesome work! Also interested in trying out stack and I had the following questions related to this notebook:

• It seems like ctrl_adata and all_context_adata refer to the same subsets?
• Also, ctrl_adata is not used in the following cells so was wondering what it means to use a synthetic control within stack-generation?
• For both stack-large and stack-large-aligned are we limited to predicting on the set of 15012 genes specified in the genelist file?

Thanks!