Excluding 1st dimension when computing SEACells on ATAC data

[kjtreese]

Hi all! Hope you're well. Thank you so much for providing this great tool :)

This may be a misunderstanding on my part about how the graph-based algorithm works, but typically when doing analyses with ATAC data, the 1st dimension should be excluded since that 1st dimension often correlations with technical variation.. I don't see where I can specify this when computing SEACells on my ATAC data. I'm seeing some technical variation in some SEACells that are grouping 1 or 2 cells from over 100 clusters in our data, and overall it seems like in these cases the cells have lowish fragment counts. I'm wondering if it's possible to exclude the 1st dimension when building the SEACells which might reduce this phenomenon? I did do this when generating my UMAP in the Anndata object, but I'm not sure that ensures the algorithm also excludes it. Thanks in advance!