Removing batch effects

[ConDem94]

Hello,

I am planning on identifying de novo clusters based on sub-cluster cell type proportions across 90+ samples. However there are clear differences between the studies in the abundance of cells, not the clustering as I have performed integration (scvi) and clusters separate well based on biological variation.

Is there a way to use this package to identify batch-adjusted proportions in order to perform my downstream clustering?

Best wishes,
Constantinos