Batch effect or biological difference

Question

[cynthier]

Hi, I have scRNA data from two stages, and they mixed well in PCA. However, the same cell types are separated in UMAP (with only one cell type mixed). I am thinking whether there is batch effect or the differences are caused by biological significance. Any suggestion is welcomed, thank you!

[ATpoint]

PC1 is probably sequencing depth -- it's common that much variation in single-cell comes from the highly uneven per-cell coverage. Beyond that, I would think about integrating this dataset for the sake of having a clustering landscape driven by celltypes rather than condition. Otherwise it is hard to really match the same cluster between conditions 1-to-1.