Lineage commitment of dermal fibroblast progenitors is controlled by Kdm6b-mediated chromatin demethylation.
The main results reproduced: recomputed values matched the published ones within tolerance.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓Same input data as the authors
- ✓Reported values were directly comparable
- ✓Reported values are derivable from the shared data
- ✓Any deviation was negligible
- ✓The central claim held under reproduction
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡Overall, the reproduction showed a material discrepancy
A 0–100 reproducibility-quality score from the per-question grades, shown as a z-score: standard deviations above (+) or below (−) the mean of comparable assessments.
▸Reproduction agent’s raw note
PRELIMINARY (fresh re-run «job» in flight) — will be confirmed when outputs return. DESCRIBED WELL ENOUGH TO REPRODUCE 1:1. The paper ships its OWN analysis code (DriskellLab GitHub, notebook scRNA_Time_Figure1.ipynb) using scanpy 1.9.1 + scanorama 1.7.4 + scvelo on the deposited GSE227257 looms (E14.5/E17.5/P5 mouse whole skin). Two prior independent «our HPC» runs («job», 2218143) reproduced the deterministic pipeline outputs EXACTLY: 43,740 cells pre-QC (= paper headline), 35,935 after QC, E14 10,320 / E17 7,183 / P5 18,432, HVG 4,000, scanorama (35935,50), leiden res=0.1 = exactly 15 cell-type clusters with near-identical markers; dermal-fibroblast lineage (clusters 0+2+5) = 17,538 cells (vs notebook 17,572 / paper 17,356). FLAGS: (1) paper '43,740' is PRE-QC, not the analyzed 35,935; (2) paper's printed 17,356 fibroblasts != authors' own committed-code output 17,572 (delta 216); (3) code normalizes to per-cell median not the '10,000 reads' in Methods; (4) notebook cell-30 fibroblast selection internally inconsistent. NOT ATTEMPTED (out of scope): RNA-velocity/PAGA, Kdm6b-KO scRNA-seq Fig 5, scATAC/ArchR, all wet-lab.
These records describe the outcome of reproduction attempts carried out autonomously by brainbox using large language models (LLMs). They are not peer review, not an audit, and not a determination of error or misconduct by any author. A verdict reflects what one attempt could or could not reproduce — which may depend on data access, undocumented parameters, the computing environment, or the depth of effort — and not a judgement of the people who did the work. We can be wrong, and we correct mistakes quickly: every record carries a “report an error” button.
Assessment versions
Every reproduction run is kept as an immutable version — anchored to the data as it stood, with a tamper-evident chain hash. A rerun (e.g. after an author updates a deposit) adds a new version; the previous one stays on record.
-
v1 current initial assessment Score 91assessed: 2026-06-16 ⛓ f9f29501a7aa
✎ I am an author of this paper
Updated or fixed a deposit, or is there an erratum? Ask us to re-run the metrics. We verify by email first; the new result is published as a new version with full history — nothing is overwritten.
Provenance — full disclosure
When this reproduction was carried out, which methodology version was used, and by whom — so the record can be audited and checked independently.
- Reproduced
- 2026-06-22
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-16no human curator yet
- Last updated
- 2026-08-05
Provisional, curator- or AI-assessed, and independently checkable. A reproduction outcome states what one attempt could reproduce — not a judgement of the authors.
Deep full-text extraction
Model: sonnetThe differentiation potential of dermal fibroblast progenitors (DFPs) is governed by chromatin accessibility and epigenetic regulation, specifically that chromatin derepression of DFPs is modulated by the H3K27me3 demethylase Kdm6b/Jmjd3.
- ★ E14.5 DFPs have a repressed transcriptional profile marked by high H3K27me3 and inaccessible chromatin at lineage-specific genes finding
- ★ Despite being multipotent, E14.5 DFPs fail to reform functional skin/hair follicles in chamber grafting assays finding
- ★ Chromatin accessibility increases specifically at lineage driver genes as DFPs differentiate from E14.5 to E18.5 finding
- ★ Kdm6b/Jmjd3-mediated removal of H3K27me3 derepresses chromatin to enable DFP lineage commitment mechanism
- ★ Dermal fibroblast-specific deletion of Kdm6b/Jmjd3 in mice causes adipocyte compartment ablation and inhibits mature dermal papilla function finding
- A publicly available multiomics search tool/dataset was generated (skinregeneration.org) resource
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| single-cell RNA-seq | whole murine skin (E14.5, E18.5, P5) | none | cell type/lineage identity, transcriptomic clustering (PAGA), RNA velocity trajectories | — |
| ex vivo chamber grafting assay | dermal + epidermal cells from E14.5, E18.5, P5 (ROSA-CMV-tdTomato) mice grafted onto Foxn1-/- host mice | transplantation/grafting | hair follicle formation and functional skin reformation | — |
| single-cell ATAC-seq | murine dermal fibroblasts at E14.5 and E18.5 | none | chromatin accessibility, ATAC peak number per cell, differential peak analysis | — |
| ChIP-seq (H3K27me3) | murine dermal fibroblasts, E14.5 vs E18.5 | none | genome-wide H3K27me3 occupancy/levels | — |
| genetic ablation (conditional knockout) with scRNA-seq, ChIP-seq, and allografting | dermal fibroblast-specific Kdm6b/Jmjd3 knockout mice | KO | adipocyte compartment formation, dermal papilla maturation/function | — |
- – E14.5 DFPs failed to regenerate hair follicles in chamber grafting assays while E18.5 and P5 fibroblasts successfully reformed functional hair follicles
- ▲ Total ATAC peak number increased from E14.5 to E18.5 fibroblasts 6,200 peaks (E14.5) vs 13,566 peaks (E18.5)
- ▲ ATAC peaks per cell increased substantially from E14.5 to E18.5
- ▼ Dermal fibroblast-specific Kdm6b/Jmjd3 deletion resulted in adipocyte compartment ablation and inhibition of mature dermal papilla functions
- – scRNA-seq captured 43,740 total cells across E14.5/E18.5/P5, of which 17,356 were classified as dermal fibroblasts
- – 10,913 fibroblasts (4,187 from E14.5, 6,006 from E18.5) passed QC for scATAC-seq downstream analysis
- count 43,740 total cells sequenced (scRNA-seq across E14.5, E18.5, P5 whole skin)
- count 17,356 cells classified as dermal fibroblasts (subset of total scRNA-seq cells)
- count 6,200 differential peaks in E14.5 vs 13,566 in E18.5 fibroblasts (Wilcoxon differential peak analysis, scATAC-seq)
- count 10,913 fibroblasts total (4,187 E14.5 and 6,006 E18.5) (cells used for scATAC-seq downstream analysis after QC)
Statistical methods review
Model: sonnetA neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.
This study employed multimodal single-cell approaches (scRNA-seq and scATAC-seq) combined with ChIP-seq, genetic knockout, and ex vivo chamber grafting assays to characterize dermal fibroblast progenitor differentiation. Computational methods including PAGA-based integration, RNA velocity, and pseudotime trajectory analysis were used to infer lineage transitions across three developmental timepoints (E14.5, E18.5, P5). Differential chromatin accessibility between timepoints was assessed with Wilcoxon testing, and grafting outcomes were quantified descriptively. The text provided is truncated before the full statistical reporting for later results sections.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Wilcoxon rank-sum test (differential peak analysis) | scATAC-seq differential peak analysis comparing E14.5 vs E18.5 fibroblasts | 10,913 fibroblasts (4,187 E14.5; 6,006 E18.5) | not stated |
| RNA velocity (scVelo dynamical model) | Trajectory inference from DFPs to fibroblast lineages in PAGA-embedded scRNA-seq | 17,356 dermal fibroblasts (from 43,740 total cells) | not stated |
| Unsupervised clustering with cross-modality label transfer (Seurat) | scATAC-seq cluster identification via integration with scRNA-seq reference | 10,913 fibroblasts | not stated |
| Quantification of hair formations (specific test not stated) | Chamber grafting assay comparing E14.5, E18.5, and P5 fibroblast populations (Fig 1J) | — | not stated |
-
Differential chromatin accessibility between E14.5 and E18.5 was assessed with Wilcoxon rank-sum testing applied per peak across cells.↳ Could also: Negative-binomial-based methods such as DESeq2, edgeR, or ArchR's built-in pseudo-bulk approach aggregate cells into per-sample pseudo-bulk counts before testing. — Pseudo-bulk approaches account for within-sample correlation (cells from the same animal are not independent observations), which single-cell Wilcoxon tests do not model; they also allow biological replication to be incorporated explicitly, reducing inflated false-positive rates that can arise when individual cells are treated as independent units.
-
Lineage trajectories were inferred using RNA velocity (dynamical model) to assign directional flow between cell states.↳ Could also: Diffusion pseudotime (DPT) or Monocle3's principal-graph-based pseudotime could also order cells along a differentiation axis without relying on spliced/unspliced mRNA kinetics. — Velocity inference assumes specific kinetic models of splicing and can be sensitive to batch effects and genes with complex kinetics (as the authors themselves note for MURK genes); pseudotime approaches that use manifold geometry alone provide a complementary, assumption-lighter trajectory estimate and can serve as cross-validation.
-
Grafting assay outcomes (hair follicle count per graft) were quantified and plotted (Fig 1J) without a stated statistical test or reported p-value.↳ Could also: A one-way ANOVA followed by a post-hoc correction (e.g., Tukey HSD) or a Kruskal-Wallis test with Dunn's post-hoc for non-normal small-n data would formally compare all three timepoint groups while controlling family-wise error. — Stating the test, p-values, and a measure of spread (SD or IQR) would allow readers to assess the magnitude and reliability of differences between groups, and is standard practice for quantitative functional assays.
-
scRNA-seq and scATAC-seq were each performed at the timepoints of interest without explicit statement of the number of independent biological replicates contributing to each timepoint.↳ Could also: Collecting and separately processing samples from at least two or three independent animals per timepoint, then treating animal as a random effect or using pseudo-bulk aggregation, would allow biological variability to be estimated and reported. — A single pooled sample per timepoint conflates biological with technical variation; multiple biological replicates are needed to generalize findings beyond the specific animals profiled and are increasingly expected by journals for single-cell genomics studies.
-
Chromatin accessibility differences between E14.5 and E18.5 were described by the raw number of differentially accessible peaks (6,200 vs 13,566).↳ Could also: Reporting fold-change or a normalized enrichment score alongside peak counts, and providing a volcano plot or MA plot, would also convey the effect size and directionality of accessibility changes. — Peak counts alone do not indicate the magnitude of accessibility change; effect-size measures help distinguish biologically meaningful from statistically significant but small differences, especially given the large cell numbers involved.
-
Cluster identity in scATAC-seq was assigned by label transfer from the scRNA-seq reference (Seurat).↳ Could also: Gene-activity score matrices derived from chromatin accessibility (e.g., via Cicero or ArchR) followed by independent marker-gene scoring would provide an orthogonal cluster-annotation strategy. — Label transfer propagates any mis-classification from the RNA reference; a parallel annotation based on accessibility-derived gene activity serves as cross-validation and is especially useful when RNA and chromatin data do not cluster identically.
Citation network
Where this publication sits in the reproducibility-weighted citation graph — what it is built on, and what is built on it. Citation data from OpenAlex.
No assessed neighbours yet — the network grows as more papers are assessed.
Data lineage
The datasets this paper uses (text-mined from the full text via Europe PMC), and which other assessed papers stand on the same data. A shared dataset is a factual link — not a judgement.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-37602956
Paper: Phan et al., "Lineage commitment of dermal fibroblast progenitors is controlled by Kdm6b-mediated chromatin demethylation." EMBO J 2023. PMID 37602956 · PMCID PMC10548174 · DOI 10.15252/embj.2023113880
Code artifacts
- BRIEF lists
github.com/brianhie/scanorama(third-party integration tool — P16 valid). - The paper's own analysis code is at
github.com/DriskellLab/Lineage-Commitment-of-Dermal-Fibroblast-Progenitors-is-Mediated-by-Chromatin-De-repression(commit pushed 2023-06-16). It ships Jupyter notebooks + R/ArchR Rmd files. The relevant notebook for GSE227257 isscRNA_Time_Figure1.ipynb— it reads exactly the 3 deposited loom files and runs scanpy + scanorama + scvelo. We reproduce by running that committed pipeline on the deposited data.
Data
- GEO GSE227257: 3 scRNA-seq samples (10x 3' v2), mouse whole skin.
- GSM7093921
e14.loom(E14.5) - GSM7093922
e17.loom(E17.5) [paper text sometimes writes "E18.5"] - GSM7093923
p05.loom(P5) loom files generated by Velocyto (spliced/unspliced layers present).
- GSM7093921
IN SCOPE (pipeline-derived, reproduced here)
Pipeline = CellRanger v6 → Velocyto loom (upstream, provided as deposited data) → scanpy QC/normalize/HVG → scanorama batch-integration by Time → neighbors/UMAP → leiden clustering → marker genes. We re-run from the deposited loom files (i.e. from the Velocyto output onward — alignment/quantification is upstream of the deposited data and not re-run).
Comparable reported values:
| id | result | reported | type |
|---|---|---|---|
| C1 | total cells captured (3 timepoints, pre-QC concat) | 43,740 | deterministic |
| C2 | total cells after QC filter | 35,935 (notebook output) | deterministic |
| C3 | E14 cells after QC | 10,320 (notebook output) | deterministic |
| C4 | dermal fibroblast cells | 17,356 (paper) / 17,572 (notebook) | semi-stochastic |
| C5 | leiden res=0.1 cell-type clusters & markers | ~15 types, Fig 1 | stochastic |
| C6 | DFP markers (Upper: Crabp1,Cav1,Nkd1,Lef1; Lower: Thbs1/Mfap5,Ptn) | Results/Fig 2 | qualitative |
OUT OF SCOPE (not attempted)
- scATAC-seq / ArchR analysis (separate accessions GSE227262/GSE233161/GSE227256, separate Rmd notebooks) — not the BRIEF's accession.
- Kdm6b KO scRNA-seq (Figure 5 notebooks) — separate experiment.
- RNA-velocity dynamical latent-time, PAGA trajectory (downstream of, and qualitatively dependent on, the stochastic embedding; not a single comparable number).
- All wet-lab results (IF/IHC, ChIP, lineage tracing, qPCR).
Known reproducibility hazards (flagged for the human auditor)
- Paper "43,740" is the PRE-QC concat count, not the analyzed count. The committed notebook drops to 35,935 after QC. The paper text presents 43,740 as "total"; that equals the raw concatenation, exact.
- Fibroblast count: paper says 17,356, the committed notebook output says 17,572 (Δ=216). The shipped code does not reproduce the paper's printed value exactly — possible-discrepancy note.
normalize_total()in the code has NO target_sum → normalizes to the per-cell median, whereas the Methods text says "normalized to 10,000 reads". We follow the committed CODE (faithful to what was run).- Notebook cell 30 fibroblast selection
leiden.isin(['0','1','3','5','7','10','22'])is internally inconsistent with the res=0.1 clustering (cluster '22' does not exist at res=0.1; labels 1/3 are keratinocytes in the same notebook's map) → the notebook was partially re-run at a different resolution. We therefore identify fibroblast clusters by canonical markers instead and report the count. - scanorama (randomized SVD) + leiden are not byte-deterministic across versions; exact cluster IDs/counts may differ. Cell-count claims C1–C3 are deterministic.
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
An automated assessment. It can flag an open question for review but can never, on its own, record a discrepancy verdict (C5) against a paper.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
The Figure-1 whole-skin scRNA-seq pipeline reproduces essentially 1:1 on the authors' own deposited GSE227257 looms: pre-QC 43,740, post-QC 35,935, per-timepoint counts, and 15 leiden clusters with matching markers all hit exactly, and the Upper/Lower DFP marker split (Fig 2) reproduces qualitatively. Deviations are minor and on the technical/stochastic + authors-reporting side, not the data side: the fibroblast count lands within ~1% (stochastic clustering), the paper's printed 17,356 disagrees with its own code's 17,572 (Δ216), and the 43,740 headline is the pre-QC number while the analyzed 35,935 is never stated. Nothing is fabrication-suspect — every value is derivable from shared data. Note the paper's actual mechanistic claim (Kdm6b KO, scATAC) was out of scope and untested, so q7 reflects only the reproduced atlas/DFP claims.
Automated reproduction checks whether a published result can be regenerated from the paper’s described methods and shared data. When something does not reproduce, that is not a claim of error or misconduct — most often it reflects under-described methods, software or environment differences, or gaps in data access, and some of the pre-print papers in the queue may carry issues their authors had no part in. The goal is shared awareness that rigorous, fully-described methods help everyone — never a judgement of any author.
Are you an author? We would genuinely like to hear from you — to clarify the record, add data or code, re-run the pipeline after an accession update, and publish your response right next to the assessment. Everything here is open and auditable.
🚩 Report an error in this record
Spotted something wrong — a verdict you’d contest, a data or value error, or a private detail that slipped through? Tell us, with a short justification. Authors and readers are equally welcome to write in; we review every report.
Prefer email, or the form below not working? Contact us at support@doesitreproduce.com.
Reproduction footprint
claude-opus-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.