I'm working with ATACSeq data from multiple tissue/cell types. The data is binned in 1 megabase bins. I'd like to identify the bins that are "highly variable" across the different tissue types. I can't seem to find a tool to do this, like there exists for identifying highly variables genes in a gene expression setting (e.g Scanpy's scanpy.pp.highly_variable_genes). I was wondering if I could use the gene selection tools for my data, and if there are any other suggestions?


1 Answer 1


I don't think is a good idea to apply the gene selection method on the ATAC-seq genomic regions because of the high sparsity of ATAC-seq data.

A common way is to select the most frequent regions (like the Signac does), or to discard the house-keeping regions (like the SnapATAC does).

And you can also use the episcanpy, a epigenomics version of scanpy.

  • $\begingroup$ Thanks for your answer. I guess in my case the cells are grouped per cell type (the data is aggregated per cell type), so sparsity shouldn't be an issue. Given this, is it still inappropriate to use gene selection methods? $\endgroup$ Commented Aug 16, 2022 at 16:57
  • $\begingroup$ If you think the sparsity is not an issue for your data, then scanpy.pp.highly_variable_genes may be applicable. Just be aware that the default thresholds for mean and disp are for single-cell RNA-seq data. $\endgroup$
    – Janus
    Commented Aug 17, 2022 at 12:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.