H3K27 and H3K9 methylation mask potential CTCF binding sites to maintain 3D genome integrity.
Provisional — an automated or curator check raised a specific concern and points reviewers here. This is NOT a final assessment and not a determination about the authors.
The main results reproduced, with only marginal, non-material deviations.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- Nothing in this column.
- 🔴Could not use the authors’ exact input data
- 🔴Reported values were only indirectly comparable
- 🟡A deviation arose in the data or preprocessing
- 🔴A deviation was attributed to the published material
- 🟡Reported values were not (fully) derivable from the shared data
- 🟡The deviation was non-trivial in magnitude
- 🟡The central claim did not (fully) hold under reproduction
- 🟡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
DESCRIBED WELL ENOUGH? Pipeline YES (Methods name Bowtie2/mm10, nf-core/chipseq v2.0.0, MACS2, DiffBind, 4DN Hi-C v43, Juicer KR 250kb, GENOVA 40kb); data only PARTIAL for the headline claims. RESULT = 1:1 for the in-scope deposited processed outputs. Regenerated all 3 deposited histone-ChIP log2ratio bigWigs from the deposited raw fastq via the stated tool chain (Bowtie2 2.5.4 -> UCSC mm10 -> samtools dedup -> deepTools bamCompare --operation log2 --binSize 50 --extendReads ChIP-over-matched-input) and compared genome-wide (multiBigwigSummary 10kb bins -> Spearman/Pearson). Agreement vs the authors' deposited tracks: RU-CHIP-1 (GSM9003705) 0.9920/0.9915; RU-CHIP-2 (GSM9003706) 0.9893/0.9884; RU-CHIP-3 (GSM9003699, H3K9me3) 0.9907/0.9925 - all EXACT-grade (>0.95). The deposited ChIP-seq processing is faithfully reproducible across both marks and both replicates. KEY AUDITABLE FINDING (unchanged): the named accession GSE297986/PRJNA1267077 deposits ONLY 12 histone ChIP-seq runs - NO CTCF ChIP-seq and NO Hi-C - so the paper's headline pipeline numbers (42,451 WT CTCF peaks; 16,521/13,665 5KO+DS DE peaks Fig 1B; Hi-C insulation counts Suppl Fig S3C) are NOT reproducible from the named accession (inputs external/reused per Suppl Table S4). Recorded as no_data_accession, NOT a fabrication flag (data plausibly exists under other accessions). NOT attempted: anything needing CTCF or Hi-C data (out of scope), and the eight-cell H3K27me3 domain analysis (external embryo data). Compute: SLURM «job» (node n096, 1h39m, exit 0). Debug note: prior jobs failed on a set -u 'srr: unbound variable' from local srr=$1 bam="...${srr}..." (same-line self-reference), masked by a $(align) command-substitution subshell + SLURM stdout block-buffering; fixed by splitting the local, running align as a statement, and live tee+stdbuf logging. Also hit a transient shared-account «infra»+home quota that cleared after janitor reclaim.
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 30assessed: 2026-06-14 ⛓ 4e4c377d22aa
✎ 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
- 2026-06-23
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15no 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 paper tests whether repressive chromatin modifications (H3K9 and H3K27 methylation) control where CTCF can bind the genome, thereby maintaining proper 3D genome organization during development.
- ★ H3K9 and H3K27 methylation regulate CTCF binding at distinct genomic regions, and their simultaneous loss induces drastic changes in CTCF binding finding
- ★ H3K9 methyltransferases (SETDB1, SUV39H1/2, EHMT1/2) primarily function to prevent CTCF binding, acting redundantly or independently depending on genomic region finding
- ★ EZH1/2 inhibition (DS3201) predominantly decreases CTCF binding, in contrast to H3K9 methyltransferase loss which predominantly increases it finding
- ★ H3K9 and H3K27 methylation act redundantly to repress CTCF binding at transposable elements (e.g., SINE B3, L1, ERV/IAPLTR2), with H3K27me3 redistributing onto transposons after loss of H3K9 methylation mechanism
- ★ Changes in CTCF DE peaks are associated with changes in nearby gene expression and 3D genome architecture/compartments finding
- ★ CTCF sites repressed by H3K9 methylation alone are bound by CTCF in early mouse embryos, whereas sites repressed by both H3K9 and H3K27 methylation remain inaccessible due to early embryo-specific H3K27 methylation finding
- DiffBind was used to identify differentially enriched CTCF ChIP-seq peaks (FDR<0.05), classified into 5 classes by dependency on Setdb1, Suv39h1/2, Ehmt1/2, and Ezh1/2 method
- A complete H3K9 methyltransferase-deficient iMEF line (5KO) combined with EZH1/2 inhibitor DS3201 was used as a system to simultaneously deplete H3K9 and H3K27 methylation resource
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| CTCF ChIP-seq | iMEFs (WT, Setdb1 KO, TKO, 5KO) ± DS3201 | KO/drug (EZH1/2 inhibitor) | CTCF binding, differentially enriched (DE) peaks | — |
| H3K9me3 ChIP-seq (reanalysis) | iMEFs WT, Setdb1 KO | KO | H3K9me3 enrichment at DE peaks/SINE B3/transposons | — |
| H3K9me2 ChIP-seq (reanalysis) | iMEFs WT, Setdb1 KO, TKO | KO | H3K9me2 enrichment at DE peaks/SINE B3 | — |
| H3K27me3 ChIP-seq/profiling (reanalysis) | iMEFs WT, TKO, 5KO ± DS3201 | KO/drug | H3K27me3 enrichment at DE peaks/transposons | — |
| RNA-seq (reanalysis) | iMEFs (Setdb1 KO, TKO, 5KO, 5KO+DS) | KO/drug | differentially expressed genes near CTCF DE peaks | — |
| Hi-C (reanalysis) | iMEFs KO lines | KO | 3D genome architecture, A/B compartments (PC1), TADs | — |
| Whole-genome DNA methylation reanalysis | iMEFs WT vs 5KO | KO | DNA methylation status at Increased CTCF binding sites | — |
| CTCF binding/H3K27 methylation assessment (reanalysis) | early mouse embryos | none (developmental stage) | CTCF occupancy and H3K27 methylation at H3K9/H3K27-repressed sites | — |
- – 5KO+DS cells show 16,521 Increased and 13,665 Decreased CTCF DE peaks, each ~35% of the average WT peak number 16,521 increased / 13,665 decreased peaks (~35% of avg WT 42,451 peaks)
- ▲ H3K9 methyltransferase-deficient cells (Setdb1 KO, TKO, 5KO) show more Increased than Decreased DE peaks
- ▼ DS (EZH1/2 inhibitor) treatment produces more Decreased than Increased DE peaks in all cell types tested
- ▼ Increased Class1 (Setdb1-dependent) peaks show loss of H3K9me2 and H3K9me3 upon Setdb1 KO adj. P=1.3e-176 (H3K9me2), adj. P≈0 (H3K9me3)
- – SINE B3 elements are enriched in Increased Class 1, 2, and 3, but only ~9% of all SINE B3 copies overlap CTCF binding sites ~9%
- ▲ H3K27me3 redistributes onto transposons (IAPLTR2_Mm, L1Fd_2) enriched in Increased Class 5 under 5KO conditions, coinciding with loss of H3K9me3
- – Genes upregulated in 5KO/5KO+DS are enriched around Increased Class3/Class5 peaks; genes downregulated are enriched around Decreased Class1/2/3/5 peaks
- – Increased Class 2, 3, and 5 peaks are enriched in intergenic regions and B compartments relative to other classes P<2.2e-16
- count 16,521 increased DE peaks; 13,665 decreased DE peaks (5KO+DS CTCF ChIP-seq DE peaks vs average WT peak count of 42,451)
- pvalue adj. P = 1.3×10^-176 (H3K9me2 decrease at Increased Class1 peaks in Setdb1 KO)
- pvalue adj. P ≈ 0 (H3K9me3 decrease at Increased Class1 peaks in Setdb1 KO)
- pvalue adj. P = 2.7×10^-24 (H3K9me3 increase at Increased Class2 peaks in Setdb1 KO)
- pvalue adj. P = 8.5×10^-260 (H3K9me2 decrease at Increased Class2 peaks in TKO)
- pvalue adj. P = 2.3×10^-70 (H3K9me3 decrease at Increased Class3 peaks in Setdb1 KO)
- pvalue adj. P < 1.3×10^-56 (H3K27me3 increase at Increased Class2 and Class5 peaks in TKO/5KO)
- other ~9% (fraction of SINE B3 copies overlapping CTCF binding sites)
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 ChIP-seq/RNA-seq/Hi-C study characterized how H3K9 and H3K27 methylation regulate CTCF binding across a series of H3K9 methyltransferase knockout and EZH1/2-inhibitor-treated immortalized mouse embryonic fibroblast lines (8+ conditions). Differentially enriched CTCF peaks were called with DiffBind (FDR < 0.05) comparing each condition to wild-type. Chromatin modification enrichment at peak classes was compared with t-tests corrected by the Benjamini–Hochberg method, and genomic distribution differences were assessed with proportion tests. Results were reported as high-precision adjusted P-values; the paper notes that the full statistical output is in Supplemental Table S1.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| DiffBind differential enrichment analysis (FDR < 0.05) | Identification of increased and decreased CTCF ChIP-seq peaks in each KO/DS condition vs. wild-type (Fig. 1B, 1D and DE peak classification) | 2 biological replicates per condition; two independent clones (#14 and #55) used for 5KO | not stated |
| t-test with Benjamini–Hochberg (BH) multiple testing correction | Comparison of H3K9me2, H3K9me3, and H3K27me3 enrichment at DE peak classes across conditions (Fig. 1E; Supplemental Fig. S1E–G; Supplemental Table S1) | number of peaks per class (not explicitly stated in excerpt) | not stated |
| Proportion test (prop.test) | Comparison of intergenic vs. genic genomic distribution among DE peak classes (Supplemental Fig. S1C); P < 2.2 × 10⁻¹⁶ reported | null | not stated |
| Repeat/transposon enrichment analysis with volcano plot display (adj. P < 0.01 threshold; specific test not named in excerpt) | Enrichment of repeat types in each DE peak class compared to consensus peaks (Fig. 2A; Supplemental Table S2) | null | not stated |
-
CTCF ChIP-seq differential binding was assessed using 2 biological replicates per condition↳ Could also: 3 or more biological replicates per condition — Increasing replicates improves variance estimation within DiffBind/DESeq2 peak-calling models and statistical power for differential enrichment, which is particularly relevant when multiple hierarchical KO conditions are compared sequentially
-
Chromatin modification enrichment differences at DE peak classes were compared with t-tests↳ Could also: Wilcoxon rank-sum test or a permutation-based test — ChIP-seq enrichment scores at genomic regions are typically right-skewed; non-parametric tests make fewer distributional assumptions and are commonly applied to such data as a complementary or primary approach
-
Each KO/DS condition was compared to wild-type individually via separate DiffBind analyses↳ Could also: A joint multi-group model (e.g., DESeq2 likelihood-ratio test or edgeR GLM across all conditions simultaneously with a design matrix encoding the KO hierarchy) — A single joint model can exploit the hierarchical structure of the KO series, improve power through shared dispersion estimation, and directly test interaction effects between H3K9 and H3K27 deficiency without inflating the number of pairwise comparisons
-
Aggregate ChIP-seq enrichment meta-profiles are presented without dispersion measures around the mean signal↳ Could also: Display 95% confidence interval or standard deviation ribbons on enrichment profiles — Showing replicate-level spread directly in the figure allows readers to visually assess between-replicate variability alongside the reported adjusted P-values, which is standard in many genomics publications
-
Repeat/transposon enrichment in DE peak classes was evaluated relative to consensus peaks using an adj. P threshold, with the underlying test not named in the excerpt↳ Could also: Permutation-based enrichment test (e.g., bedtools shuffle, HOMER repeat enrichment, or GAT) — Permutation-based approaches directly control for genomic size, mappability, and repeat density biases by generating empirical null distributions, providing a complementary validation of enrichment results
-
Genomic distribution differences among DE peak classes were assessed with prop.test, apparently applied per class, without explicit correction across the multiple classes↳ Could also: Chi-squared test of independence across all classes simultaneously, followed by post-hoc pairwise comparisons with FDR correction — An omnibus test followed by corrected pairwise comparisons controls the family-wise error rate across the set of classes being compared, whereas repeated uncorrected proportion tests increase the nominal Type I error rate
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.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-40764058
Paper: Fukuda K, Shimura C, Shinkai Y. H3K27 and H3K9 methylation mask potential CTCF binding sites to maintain 3D genome integrity. Genome Res 2025. PMID 40764058 · PMCID PMC12487818 · DOI 10.1101/gr.280732.125
Named code artifact (manifest): https://github.com/4dn-dcic/docker-4dn-hic (the 4DN Hi-C Docker pipeline, v43 — a third-party tool; P16-valid). Data accession (manifest): GEO GSE297986 (SRA BioProject PRJNA1267077).
What the deposited data actually is
GSE297986 / PRJNA1267077 contains exactly 12 paired-end ChIP-seq runs (verified via ENA filereport), all histone modifications in immortalized mouse embryonic fibroblasts (MEFs):
| GSM | run | assay | genotype |
|---|---|---|---|
| GSM9003699 | SRR33679346 | H3K9me3 | Setdb1 KO |
| GSM9003700 | SRR33679345 | H3K9me3 | TKO |
| GSM9003701 | SRR33679344 | H3K9me2 | TKO rep1 |
| GSM9003702 | SRR33679343 | H3K9me2 | TKO rep2 |
| GSM9003703 | SRR33679342 | H3K9me2 | 5KO rep1 |
| GSM9003704 | SRR33679341 | H3K9me2 | 5KO rep2 |
| GSM9003705 | SRR33679340 | H3K27me3 | TKO rep1 |
| GSM9003706 | SRR33679339 | H3K27me3 | TKO rep2 |
| GSM9003707 | SRR33679338 | Input | Setdb1 KO |
| GSM9003708 | SRR33679337 | Input | TKO rep1 |
| GSM9003709 | SRR33679336 | Input | TKO rep2 |
| GSM9113789 | SRR34571435 | H3K9me2 | Setdb1 KO |
Deposited processed files = per-sample *.log2ratio.bw bigWig tracks
(ChIP-over-input log2 coverage). Stated pipeline (Methods): adapter trim
(Trim Galore) → Bowtie2 → mm10 → Picard dedup → bigWig; nf-core/chipseq
v2.0.0; MACS2 narrow peaks; DiffBind for differential analysis.
In scope (reproducible from the DEPOSITED data) — ATTEMPTED
RU-CHIP — regenerate the deposited histone-ChIP log2ratio bigWig from raw
fastq with the stated pipeline, and quantify agreement against the authors'
deposited track. This is a true 1:1 of a pipeline-derived processed output:
the deposited *.log2ratio.bw is the reported artifact; I rebuild it from the
deposited fastq using the described tool chain (Bowtie2/mm10 → dedup → deepTools
bamCompare --operation log2 ChIP-vs-matched-input → bigWig) and compare
genome-wide (deepTools multiBigwigSummary bins → Spearman/Pearson r).
Targets (clean ChIP / matched-input pairs that each have a deposited bigWig):
- H3K27me3 TKO rep1: ChIP SRR33679340 / Input SRR33679337 → GSM9003705
- H3K27me3 TKO rep2: ChIP SRR33679339 / Input SRR33679336 → GSM9003706
- H3K9me3 Setdb1KO : ChIP SRR33679346 / Input SRR33679338 → GSM9003699
Agreement metric: genome-wide 10-kb-bin correlation (reproduced vs deposited). Secondary descriptive: MACS2 peak counts per sample (not a paper-reported number).
Out of scope (NOT derivable from the deposited data) — NOT ATTEMPTED, with reason
The paper's headline pipeline numbers are NOT reproducible from GSE297986, because the underlying data is not deposited there (Methods/Data-availability point to "Supplemental Table S4" — external/reused accessions):
- CTCF ChIP-seq results — WT CTCF consensus peaks (42,451), 5KO+DS increased
DE peaks (16,521) / decreased (13,665), DiffBind FDR<0.05 (Fig 1B). → No CTCF
ChIP-seq run exists in PRJNA1267077.
drop_reasonfor this sub-result:no_data_accession(data external, not in named accession). - Hi-C results — A/B compartments (250-kb, Juicer KR), insulation scores (GENOVA 40-kb), insulation-change region counts (Suppl Fig S3C). → No Hi-C run in PRJNA1267077; the 4DN Hi-C pipeline (the manifest's code URL) has no deposited Hi-C input to run on. Not attempted (data external + heavy compute; this is the optional last ~20%).
- Eight-cell-stage H3K27me3 domains (6,955/26,700) — derived from external published embryo H3K27me3 data, not these MEF samples. Out of scope.
Compute plan
Single «our HPC» SLURM job (partition=std, 1 node, 32 cpus), conda env built in-job (compute nodes have internet), all data + intermediates on «infra». Small results (correlation values, peak cou
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 reproduction splits cleanly: the deposited histone tracks (GSM9003705/06/99) are reproducible in principle from deposited fastq, but the «our HPC» job (2176064) was still running at cutoff, so no correlation numbers exist yet. The headline claims — 42,451 WT CTCF peaks, 16,521/13,665 DE peaks (Fig 1B), and Hi-C insulation counts (Suppl Fig S3C) — cannot be reproduced from the named accession GSE297986, which contains only histone ChIP-seq; the CTCF and Hi-C inputs are external (Suppl Table S4). This is a data-availability gap relative to the named accession, not fabrication: the central conclusion was untested rather than refuted, so overall criticality is yellow.
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.