Can you map DNA methylation in one cell?
An optimized bisulfite protocol produced 82 single-cell methylomes (median ~1.4 million CpGs each); pooling a few dozen cells covered >90% of CpGs and a region-set method placed hematopoietic and ESC samples on treatment trajectories.
Source
Single-cell DNA methylome sequencing and bioinformatic inference of epigenomic cell-state dynamics
Study at a glance
- Design
- Animal / in-vitro — Optimized μWGBS/scWGBS plus a region-set bioinformatic method applied to K562 azacytidine, HL60 vitamin D3, and mouse ESC 2i/ATRA/EB models
- N
- N=82 · 82 single-cell methylomes, 89 four-cell methylomes, and 53 optimization/QC libraries after >80% QC pass
- Population
- Human K562, HL60, and KBM7 lines plus mouse CCE ESCs (and 32D clonal-expansion tests)
- Outcome
- Single-cell methylome coverage, clustering by cell state, and inferred epigenomic dynamics from sparse data
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
FACS-sorted 1–500 cells, ran post-bisulfite adaptor WGBS (μWGBS/scWGBS), and aggregated methylation over public region sets. Models: K562 + azacytidine, HL60 + vitamin D3, and mouse ESCs in 2i vs ATRA/embryoid-body differentiation.
What they found
>80% of libraries passed QC (82 one-cell, 89 four-cell). Human one-cell: median 4.6 million reads and 1.4 million CpGs. Combining a few dozen cells covered >90% of CpGs. K562 showed global demethylation after azacytidine; HL60 changes were subtler; ESC 2i vs serum residuals placed ATRA/EB samples in the opposite direction.
The limits
What it doesn't show
Low-coverage single-cell WGBS cannot call every CpG in one cell; cell-line models are not embryos or tumors, and azacytidine demethylation is a drug effect, not normal development.
Key terms
- WGBS
- Whole-genome bisulfite sequencing, gold-standard DNA methylation mapping.
- scWGBS / μWGBS
- This paper’s low-input/single-cell and micro-input bisulfite protocols.
- Azacytidine
- DNMT-depleting hypomethylating drug used on K562 cells.
- 2i
- Two-inhibitor naïve-pluripotency medium that remodelled ESC methylomes versus serum+LIF.
- Region-set residuals
- Bioinformatic trick: average methylation across many similar genomic regions to rescue sparse single-cell data.
- Composite methylome
- Pooled low-coverage single-cell libraries that together cover most CpGs and capture cell-to-cell variation.
Flashcards
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
Single-cell methylomes reported:
Common questions
How many single-cell methylomes?
82 (plus 89 four-cell).
Median CpG coverage per human one-cell?
About 1.4 million CpGs from 4.6 million reads.
Genome coverage when pooling?
>90% of CpGs with a few dozen cells.
K562 treatment?
1 μM 5-azacytidine; global methylation fell in most cells.
More on Epigenetics