Question: Can I do a shallow sequencing run for ATAC?
Answer: In terms of a shallow sequencing run, please keep in mind that sequencing reads/depth affect peak-calling (you need coverage to determine what a peak is), which in turn affects cell-calling (you need peaks to call cells). However, we've found that sequencing 5-10k reads/cell may be enough to give you a rough estimate of how many cells are present in your sample (Note that 5k reads/cell underestimates cell count). At this coverage, clustering patterns also start to become more defined.
Although you can begin to see insert size distribution and TSS enrichment scores with a shallow sequencing analysis, other web summary metrics will likely be highly inaccurate. This may be helpful to determine sample quality in some cases, but not always.
ATAC needs more depth than other assays (Gene Expression) for peak calling and cell calling to work correctly. Sequencing less than 5k reads/cell will not provide enough data for the algorithm to work correctly.