Hi all, I have a multiplexed 10x Multiome (snRNA + snATAC) dataset where one of the samples had a technical issue I'd like guidance on.

The original sample was clogged during loading and the Cell Ranger ARC report estimated only ~1,500 cells The experiment was repeated and recovered ~5,500 cells (a separate library / sequencing run) Other samples in my cohort have ~8,000–11,000 cells per pool The pool is multiplexed (2 donors per pool), demultiplexed via genotype using Vireo

Is it appropriate to combine the original (1,500-cell) and repeat (5,500-cell), or should I treat them as separate samples (or just use the repeat and discard the original)? If combining is OK:

Should the libraries be merged at the count matrix level (concatenate cells), or should I align fragments at the sample/donor level first? Are there batch-effect considerations between the two runs that I need to account for (e.g., the two were sequenced separately, possibly with different chemistry batch / loading conditions)?

I'm relatively new to single-cell analysis so any pointers to literature or worked examples would be hugely appreciated.