scATAC-seq time course experiment about cistopic

[cbravo93]

Hi @jk86754!

I have never tried something like that (normally we check per time point and also run a combined model; so we would use the combined model for this question), but it could work. The most critical step is how to link the regions to genes (e.g. you can aggregate regions in a certain space around the TSS + introns).

Once you have your gene-based matrix, scRNA-seq oriented methods have worked quite well in our cases (e.g. Seurat, SCENIC; in your case maybe Monocle is interesting). If you see batch effects (though it may be hard to distinguish whether it is because of the time point or the runs), you can have a try with harmony; otherwise, getting the gene-based matrix on a combined model can help.

Hope this helps!

C

[jk86754]

Thank you. I had very similar ideas; it is good to get confirmation.