Biology Atlas

Map / Open register

question · deep · partly written

How much of a chromatin assay peak set is artefact?

CUT&Tag recovers about half of ENCODE ChIP-seq histone peaks, a preprint reports signal unique to CUT&Tag at promoters, and ChIP-seq itself has known artefacts, so no assay is a clean ground truth.

The short version

In K562, CUT&Tag recovered an average 54% of ENCODE H3K27ac and H3K27me3 peaks. A preprint reports 18% of K562 H3K27me3 CUT&Tag peaks have no ChIP overlap. ChIP has its own artefacts, so neither is truth. 123

Reported, not yet replicated

The three sources more

Abbasova and colleagues benchmarked CUT&Tag against ENCODE K562 ChIP-seq, average recall 54% (standard deviation 4.99; maximum 63%), noting that the peaks found represent the strongest ENCODE peaks and are enriched in cell type-specific regulatory regions. Hu and colleagues (preprint) report that on average 18% of K562 H3K27me3 CUT&Tag peaks are unique to CUT&Tag, show promoter enrichment near transcription start sites, and a survey of 277 human H3K27me3 CUT&Tag data sets from 2024 onward showed a false signal across ubiquitously expressed genes. Teytelman and colleagues found 238 hyper-ChIPable yeast loci in highly expressed regions. 123

Reported, not yet replicated

Caveats deep

The benchmark data are one cell line and two marks. The preprint's 18% is an average across data sets and depends on peak calling choices. The yeast result is a different assay and organism. 123

Reported, not yet replicated

Observation

A benchmark of CUT&Tag for H3K27ac and H3K27me3 against ENCODE ChIP-seq in K562 cells found an average recall of 54% of ENCODE peaks (best 63%), with CUT&Tag peaks representing the strongest ENCODE peaks. A preprint reports that 18% of K562 H3K27me3 CUT&Tag peaks on average do not overlap ChIP-seq peaks and that most of 277 recent human H3K27me3 CUT&Tag data sets show enrichment at promoters of ubiquitously expressed genes attributed to open-chromatin bias. In yeast, 238 euchromatic loci were enriched for unrelated proteins in ChIP. 123

Reported, not yet replicated

Hypotheses

CUT&Tag misses or adds peaks through assay bias

Tn5 preference for accessible chromatin adds false signal at active promoters, especially for repressive marks, and recall is limited. 21

Predicts. CUT&Tag-only H3K27me3 signal at accessible promoters of expressed genes, absent in ChIP-seq.

Would falsify. CUT&Tag-only signal confirmed by an orthogonal method as true H3K27me3.

Reported, not yet replicated

The ENCODE ChIP-seq yardstick is itself imperfect

Peaks missed by CUT&Tag may be noise or false signal in ChIP-seq from shearing, fixation and highly expressed loci. 13

Predicts. ChIP-seq-only peaks enriched at highly expressed regions and absent in orthogonal assays.

Would falsify. ChIP-seq-only peaks validated as true sites.

Reported, not yet replicated

What is missing

A ground truth

Neither assay is a reference for the other; the benchmark authors say it is unclear how well ENCODE data reflects the ground truth. 1

“it is unclear exactly how well ENCODE data reflects the ground truth” CUT&Tag recovers up to half of ENCODE ChIP-seq histone acetylation peaks, Discussion

Reported, not yet replicated

Peer-reviewed version of the bias result

The bias result was read in its preprint form; a journal version is reported to exist and was not opened. 2

Reported, not yet replicated

What is known

Assays agree on the strongest peaks and disagree on about half of ENCODE H3K27ac and H3K27me3 peaks in K562; bias toward open chromatin is documented for CUT&Tag; expression-dependent artefacts are documented for ChIP in yeast. 123

Reported, not yet replicated

What is not

The fraction of any assay peak set that is biology rather than artefact, and whether CUT&Tag-only signal differs from ChIP-only signal in function or conservation. 12

Reported, not yet replicated

What would settle it

Orthogonal validation of discordant peaks (for example perturbation, spike-ins or a third assay) across marks and cell types, with a peer-reviewed benchmark. 12

Our inference

Who works on it

Still open here

Linked from

Sources

  1. Abbasova, L., Urbanaviciute, P., Hu, D., et al.. CUT&Tag recovers up to half of ENCODE ChIP-seq histone acetylation peaks. Nature Communications (2025). paper; opened 2026-11-17 all records citing this
  2. Hu, S.S., Su, Z., Liu, L., et al., Zang, C.. PATTY corrects open chromatin bias for improved bulk and single-cell CUT&Tag profiling. bioRxiv (2025). preprint, not peer reviewed; opened 2026-11-17 all records citing this
  3. Teytelman, L., Thurtle, D.M., Rine, J., van Oudenaarden, A.. Highly expressed loci are vulnerable to misleading ChIP localization of multiple unrelated proteins. PNAS (2013). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (1)

This record has been changed 1 time, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.

Kind
big-open
Register
open-question
Status
partly written
Last updated
2026-11-17
Record
content/questions/how-much-of-a-chromatin-assay-peak-set-is-artefact.json