nCount_SCT varies by batch after SCTransform – normal behavior?

[AndresGithub92]

Hi, I performed SCTransform on my samples without integration, as I wanted to preserve biological differences. However, when I look at nCount_SCT, I still see clear differences across batches (attached UMAP).

Since SCTransform is supposed to normalize for sequencing depth, I'm wondering:

Is it expected that nCount_SCT still reflects batch-specific differences? If so, Why?
Could this indicate that sequencing depth differences are not fully corrected?
If so, is there a recommended way to address this?

Thanks in advance for any clarification.