-
|
I have several multiome datasets (same experiment, different timepoints so 12 total datasets). I originally merged them, created a common peak set (following this vignette: https://stuartlab.org/signac/articles/merging#creating-a-common-peak-set). I also saw where you could merge objects first, then do CallPeaks() and use the "group.by" parameter. Since my data isn't necessarily different datasets, it was all one big experiment, would it be a better approach to call peaks using the latter approach? I'm pretty new to ATACseq so any insight is appreciated. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
We demonstrated some advantages of peak calling in the Signac paper, mainly that MACS2 can do better than the cellranger peak calls (although cellranger-atac has now updated their peak calling algorithm so this may not be an issue), and that peak calling on a per-celltype basis can help to identify peaks that are unique to low-abundance cell types. If you think either of these things could help in your analysis, certainly try per-celltype peak calling with MACS2 using the CallPeaks function. |
Beta Was this translation helpful? Give feedback.
We demonstrated some advantages of peak calling in the Signac paper, mainly that MACS2 can do better than the cellranger peak calls (although cellranger-atac has now updated their peak calling algorithm so this may not be an issue), and that peak calling on a per-celltype basis can help to identify peaks that are unique to low-abundance cell types. If you think either of these things could help in your analysis, certainly try per-celltype peak calling with MACS2 using the CallPeaks function.