How to Design a Single-Cell ATAC-seq Experiment

For a comparison between two conditions in a primary tissue, plan for:

• • ≥2 donors per condition (3 is safer), with multiplexing to control batch;
• • several thousand cells per sample after QC, per cluster of interest;
• • ~25,000–50,000 read pairs per cell as a planning range, adjusted for the fraction of usable fragments;
• • nuclei, not cells, for most frozen or primary tissues;
• • an analysis plan fixed in advance: clustering resolution, marker genes, pseudobulk unit, and statistical test.

The most common failure is not depth — it is one donor per condition, which makes population-level claims untestable.

Quality is dominated by upstream handling, not by sequencing depth.

Run a small pilot with a known cell line alongside your sample: if the line does not give a clean nucleosome ladder and TSS enrichment, the sample will not either.

Cells per sample. Sparse per-cell data means you need depth in numbers. As a planning rule, aim for several thousand high-quality cells per cluster you intend to compare, not per experiment.

Multiplexing. Hashing (antibody-based or genetic) lets you pool samples and demultiplex computationally. This reduces batch effects substantially and is close to essential for cross-condition comparisons.

Replication. Biological replicates are donors or independent differentiations. Cells are never replicates — treating them as such is the single most common statistical error in this assay.

Depth affects which questions are answerable:

• • Cell-type annotation works at modest depth.
• • Differential accessibility between clusters needs enough fragments per cell in the relevant cluster.
• • Footprinting needs considerably more depth and careful fragment filtering.
• • Rare populations need both depth and cell number; a 1% population in 5,000 cells gives ~50 cells, which is not enough for differential testing.

Plan depth from the question, and state the resulting detection limit.

Write these down before library prep:

1. Clustering resolution and marker genes — the peak set you compare depends on the clusters.
2. Pseudobulk unit — donor × cluster is usually right.
3. Differential test — a count-based model at pseudobulk level, not a per-cell test.
4. Minimum cell number per cluster for inclusion.
5. Multiple testing correction across regions or peaks.
6. What counts as a result — effect size threshold as well as significance.

Decisions 1 and 6 are the ones most often made post hoc, and they are the ones that most affect the answer.

• • A reference cell line processed in parallel, for pipeline sanity.
• • Nuclei integrity check by microscopy before tagmentation.
• • Library complexity estimated from duplication and unique fragments.
• • TSS enrichment and fragment periodicity as per-sample QC; samples failing these are not rescued by depth.
• • Ambient chromatin contamination assessed from empty droplets; high background inflates co-accessibility.

• • A homogeneous cell line — bulk ATAC is cheaper and better.
• • A question about one specific element — bulk or targeted assays are more powerful.
• • Fewer than ~1,000 recoverable nuclei — consider bulk with spike-in normalization.
• • A population below ~1% without enrichment — sort or enrich first.

EDITGENE builds cell systems that follow single-cell discovery: matched cell lines or iPSC-derived models, CRISPRi/a-ready lines for element tests, and reporter lines for quantitative follow-up.

Share the tissue context and candidate elements. Contact EDITGENE to request a project consultation or quote.

Recommended cell numbers and depths are planning ranges; actual requirements depend on tissue, cluster abundance, and the specific comparison.

Single-cell accessibility data are sparse and analysis-dependent; per-cell results should be interpreted at pseudobulk level for differential claims.

• • Buenrostro JD, et al. Single-cell chromatin accessibility reveals principles of regulatory variation. Nature. 2015;523:486–490.
• • Satpathy AT, et al. Massively parallel single-cell chromatin landscapes of human immune cell development and intratumoral T cell exhaustion. Nature Biotechnology. 2019;37:925–936.
• • Stuart T, et al. Single-cell chromatin state analysis with Signac. Nature Methods. 2021;18:1333–1341.
• • Zhang K, et al. A single-cell atlas of chromatin accessibility in the human genome. Cell. 2021;184:5985–6001.