Disclaimer: This blog provides an educational summary of a preprint that has not yet been peer-reviewed. It does not represent endorsement by the study authors, SeekGene, or any journal. Performance characteristics described are observations within this specific study and shall not be generalized as universal product claims.
For Research Use Only. Not for use in diagnostic procedures.
Every tissue is a community of clones. The millions of blood cells in your body are not one interchangeable population; they descend from distinct ancestral clones, and those clones do not behave alike. A shift in blood composition can mean one clone expanded, an existing clone changed its lineage output, or a clone’s descendants remodeled their molecular state. Telling these apart has long been nearly impossible, because it requires reading a cell’s ancestry and its current state at the same time, in the same cell.
Until now, human lineage tracing has leaned on genetic markers that each carry a burden. Nuclear mutations are stable but too rare to discover cheaply at scale. Mitochondrial variants drift and confound. Copy-number changes are often absent from normal tissues. And the methods that do measure DNA methylation and RNA together have been plate-based and low-throughput, too slow and too expensive for large cohorts. The clonal structure of healthy human blood has therefore stayed largely invisible.
A new preprint from Westlake University and SeekGene, with clinical collaborators at Qilu Hospital of Shandong University and the First Affiliated Hospital, Zhejiang University School of Medicine, sets out to change that. The study introduces Droplet-MT, a droplet-based workflow that captures the transcriptome and the whole-genome methylome from the same cell at roughly 10,000 cells per library. Applying it to about 21,000 paired profiles from healthy blood and longitudinal acute myeloid leukemia, the authors reconstruct clonal ancestry from stochastic methylation epimutations rather than predefined mutations, and show how that ancestry shapes molecular state in aging and in cancer.
Which cells share a past — and how does that past shape what they become?
The scientific challenge
Linking ancestry to state, at scale
Three obstacles stood in the way. Existing joint methylome-transcriptome methods remain substantially lower-throughput and more expensive than droplet-based single-cell RNA sequencing. De novo epimutation-based reconstruction benefits from roughly 3% CpG coverage per cell, which makes discovery-depth profiling costly at cohort scale. And interpreting clonal behavior in cancer requires linking ancestry and state to genetic alterations, yet sparse methylome coverage limits reliable cell-level genotyping. Droplet-MT and its companion computational framework were designed around all three.
Why it matters
Why SeekOne™ DD Single-Cell DNA Methylation + RNA Multi-Omics matters
01
Transcriptome and methylome in the same cell
Joint profiling captures gene expression and whole-genome DNA methylation, with a median of roughly 3,000 genes and one million CpGs per cell. The transcriptome reports what a cell is doing now; the methylome keeps a more persistent record of its regulatory past.
02
Ancestry without predefined mutations
MethylTree reconstructs clonal ancestry from stochastic epimutations inherited across CpG sites. Because it needs no driver mutation, it works in normal tissues where informative variants are rare.
03
Throughput that makes cohorts tractable
Roughly 10,000 cells can be profiled per library at droplet scale, rather than the hundreds of cells typical of plate-based methods, making large cohorts and longitudinal sampling practical.
04
A built-in label-transfer framework
A high-depth reference is built, then other cells are assigned to clones at about 0.2% CpG coverage with up to 99% concordance against full-depth labels at a defined confidence cutoff. Clone-level aggregation of somatic variants adds genetic annotation without single-cell genotyping.
The workflow
A scalable workflow across healthy blood and longitudinal leukemia
The team applied Droplet-MT to healthy peripheral blood from two donors, 17 and 63 years old, and to acute myeloid leukemia, including a patient followed longitudinally from diagnosis through treatment and into ultra-shallow, low-coverage monitoring. Roughly 21,000 paired transcriptome-methylome profiles passed quality control. The workflow proceeds in stages: high-depth cells define a clonal reference, shallow cells are assigned to it by label transfer, and clone-aggregated mutation calling annotates each clone’s genetic lesions. The same analytical framework applies to healthy and malignant hematopoiesis alike.
Figure 1 | Droplet-MT jointly captures the transcriptome and whole-genome methylome from the same cell, then reconstructs clonal ancestry by MethylTree and assigns new cells by label transfer
Key findings
Four findings from the preprint
FINDING 1 clonal landscape
Healthy blood resolves into a polyclonal landscape
The two healthy donors looked nothing alike at the clonal level. The 17-year-old carried many small clones, an effective clone number of 103, whereas the 63-year-old was dominated by several expanded clones, an effective clone number of 20. That expansion was not neutral. The largest clone in the older donor made up 17% of analyzed cells but contributed 26% of all monocytes, a pronounced monocyte skew. Other clones were lymphoid-enriched, showing that clonal expansion does not push uniformly toward one fate. By reading ancestry and state in the same cell, Droplet-MT turns a vague notion of aging blood into a concrete map of which clones expanded and where their descendants went.
Figure 2 | The clonal landscape of healthy blood differs sharply between a young donor and an older donor
FINDING 2 abundance vs. state
Clonal abundance and state remodeling, separated
In acute myeloid leukemia, the framework distinguished two processes that are usually conflated. In one longitudinally sampled patient, one clone fell from 81% to 10% of cells after induction therapy, while a second clone rose from 5% to 65%. At the same time, the declining clone and a third clone shifted internally from progenitor-dominant toward greater monocyte representation, while the rising clone stayed monocyte-like. The observed rise in monocytes reflected both a change in which clones were present and a remodeling within persistent clones. Clone-level mutation annotation assigned distinct lesions to each clone: a WT1 mutation to one, an SF3B1 mutation to another, and a RUNX1 mutation spanning clones. Separating abundance from state, and both from genotype, is precisely what single-layer methods cannot do.
Figure 3 | Acute myeloid leukemia: clonal abundance and state remodeling are separated in one longitudinally sampled patient.
FINDING 3 recovery
Recovery emerges as polyclonal hematopoiesis
The same framework revealed what remission actually looks like. In a second patient, an NRAS-mutant malignant clone dominated at diagnosis but was depleted after treatment, while a recovery-associated compartment with roughly 30 effective clones and multilineage output took over. At late ultra-shallow monitoring, no NRAS mutant reads were detected among the eight callable cells of the recovery compartment. Notably, the largest recovery subclone showed myeloid bias, echoing the lineage skew seen in the older healthy donor. Recovery, the data suggest, is not a return to normal but a redistribution of clonal contributions, a process that mutation status alone cannot describe.
FINDING 4 epigenetic memory
Malignant cells keep an epigenetic memory
Across patients, malignant monocytes remained distinct from healthy and recovery-associated monocytes at both the RNA and the methylation level, even after treatment-associated remodeling. Recurrent hypermethylated regions near WT1, MIR663AHG, and PRDM8 persisted after treatment, and malignant monocytes expressed CSF1R, FLT3, and CRIP1 with enrichment of mTORC1 signaling. The two layers did not remodel in lockstep: RNA and methylation similarity changed at different rates and to different degrees depending on the clone. That asynchrony is exactly why ancestry, state, and epigenome must be measured together in the same cell.
Outlook
From a preprint to a framework for tracing human tissues
Droplet-MT offers a scalable way to ask a question that has been technically out of reach: which cells share a past, and how does that past shape what they become. In healthy blood, it maps how clonal organization differs between a young and an older donor. In leukemia, it separates clonal expansion, state remodeling, and polyclonal recovery, and it shows that malignant cells retain an epigenetic memory that RNA alone would miss. As a preprint, these findings are hypotheses awaiting peer review and prospective validation, and the two healthy donors cannot by themselves establish an age effect.
What the study does establish is a template: a droplet-scale readout that links clonal ancestry to molecular state in the same cell, at a throughput that finally matches the size of the question. If your research traces clonal dynamics in blood, aging, or cancer, this is a workflow worth watching.
Interested in SeekOne™ DD Single-Cell DNA Methylation + RNA Multi-Omics?
Shao J, Chen M, Liu Y, et al. Scalable methylome-transcriptome profiling resolves clonal organization in human hematopoiesis. bioRxiv. Posted October 11, 2026. https://doi.org/10.64898/2026.10.08.757494
