Harmony integration of integarted samples

Question

[fifty_fifty]

I need to integrate snRNA-seq data from 3 samples (S11, S12, S13) of species1 and 3 samples (S21, S22, S23) of species2. So, I have potential batch effects of library prep among samples of the same species and bio and tech batch effects between different species. Also, when I use harmony for integration, I need to specify group.by.vars which can be 'samples' or 'species'. Which option will be correct:

1. Run two separate RunHarmony for S21, S22, S23 and S21, S22, S23 and then integrate two resulting objects?
2. Integrate all samples at the same time. Can I put group.by.vars = c('sample','species') then?

What will be the best way to integrate all samples and remove batch effects associated with sample handling and inter-species heterogeniety?

Answer 1

[jared.andrews07]

You have two potential approaches. Just try each and see which seems to work better for your intended purpose. Generally, I just integrate on sample given I'm just trying to cram stuff together to annotate cell types more easily.