How to analyze a subset from multiome (ATAC + RNA-seq) data

[LuciaJim]

I have subsetted a group of clusters from my main multiome object, and I noticed that the LinkPeaks information doesn’t appear in the subset. Does it make sense to re-run LinkPeaks on the smaller object? Is it better practice to re-run the entire pipeline on the subsetted object? Thanks!

[timoast]

Yes, you can re-run LinkPeaks on the subset cells.

You should keep in mind which steps in the analysis will be affected by changing the set of cells: TF-IDF, SVD, clustering, peak-gene links, etc.

[LuciaJim]

Hi Tim, thank you for your answer.

What about motif scoring and footprinting? Can I also re-run FindMotifs, RunChromVAR, or Footprint on the subsetted object? If the answer is yes and I have a subsetted object that only contains my two groups of cells of interest, does it make sense to perform motif scoring while choosing a set of background peaks matched for GC content from the same cell types (since they are the only ones left in the object)?

I also wonder whether, when subsetting cells from the object, the peak matrix size is maintained because it is populated with zeros for those peaks that are not present in the remaining cells. In my case, I see that the number of peaks remains the same after subsetting the cells using subset(main_object, idents = "specific_metadata")

Thank you