A multiple super-enhancer region establishes inter-TAD interactions and controls Hoxa function in cranial neural crest.
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
- ✓No relevant deviation in data/preprocessing
- ✓No authors-side cause for any deviation
- ✓Reported values are derivable from the shared data
- ✓Any deviation was negligible
- ✓The central claim held under reproduction
- ✓Overall, the reproduction was clean
- Every checked point held up.
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 to reproduce 1:1. The paper's Hi-C Methods are fully specified (HiCUP 0.6.1 -> Juicer 1.6 -> hic2cool 0.8.3 -> HiCExplorer 3.7.2 hicCorrectMatrix KR / hicFindTADs @ 25 kb, mm10). The RU code artifact is the third-party tool hic2cool (P16); applying it to the paper's own data is a valid reproduction. I corrected the scaffold's data accession: the real 2023 Hi-C is GSE211901 (the scaffold listed GSE89437, the group's re-used 2017 ChIP/ATAC/RNA SuperSeries, no Hi-C). Starting from the paper's shipped Juicer .hic for PA2 E10.5 (GSM6505199, the key sample for the Hoxa2<->HIRE inter-TAD claim), I ran the exact documented downstream pipeline on «our HPC» and recovered the highlighted Hoxa+Evx1 TAD at chr6:52,150,000-52,325,000 vs the reported chr6:52,145,433-52,327,518 (both edges within a single 25 kb bin), plus a TAD boundary at ~chr6:52,150,000 sitting between the HIRE-containing domain and the Hoxa cluster -- i.e. the reported 'boundary between Hoxa2 and HIRE1/HIRE2'. hic2cool conversion, 25 kb resolution, KR normalization, and hicFindTADs (327 boundaries) all reproduced cleanly. This is a CLEAN RE-RUN (SLURM 2220359, COMPLETED) that independently reproduced the first run (2178858) byte-for-byte on inputs (PA2 .hic sha256 identical) and identically on outputs (same domain/boundary coordinates, same 2,554,330 chr6 data points / sparsity 0.071, same 327 boundaries). The Md E10.5 comparison sample (GSM6505198) again independently reproduced the identical Hoxa+Evx1 domain and the ~52.15 Mb boundary, corroborating C3/C4 a second time. NOT attempted (out of scope / hard 20%): raw-read mapping HiCUP->Juicer; ROSE super-enhancer calling; PCHi-C CHiCAGO / virtual-4C; replicate merging; all wet-lab phenotypes. No completeness claim; grades provisional pending human audit. No fabrication signal -- the compared coordinates are genuinely derivable from the shipped data via the documented tools.
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 94assessed: 2026-06-15 ⛓ bfbf98595dca
✎ 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 clusters of super-enhancers can overcome topologically associating domain (TAD) insulation to establish long-range inter-TAD regulatory interactions controlling Hoxa2 (and neighboring Hoxa gene) expression in cranial neural crest cell (CNCC) subpopulations during craniofacial development.
- ★ 2232 genome-wide putative super-enhancers (SEs) were identified in mouse cranial neural crest cell (CNCC) subpopulations resource
- ★ 147 SEs are linked to 62 positional transcription factor-coding genes (148 unique pairings) establishing CNCC positional identity finding
- ★ A multiple SE-containing region, subdivided into HIRE1 and HIRE2, establishes long-range inter-TAD interactions selectively with Hoxa2 in PA2 CNCCs, skipping the neighboring 3' TAD finding
- ★ CRISPR deletion of HIRE1 phenocopies the full homeotic Hoxa2 knockout and additionally induces PA3/PA4 CNCC-derived skeletal abnormalities correlating with Hoxa2 and Hoxa3 transcriptional downregulation finding
- ★ HIRE2 deletion alone causes no major skeletal alterations, but on a Hoxa2 haploinsufficient background it results in microtia, indicating functional redundancy finding
- HIRE1 and HIRE2 sequences are highly conserved among eutherian mammals, with lower conservation in marsupials, birds, and fish finding
- ★ Super-enhancers can overcome TAD insulation to regulate anterior Hoxa gene collinear expression in a CNCC subpopulation-specific manner mechanism
- Inter-TAD Hoxa2-HIRE interactions are visible as an asymmetrical architectural stripe on Hi-C plots and are selective for Hoxa2-expressing PA2/pinna CNCCs, absent in Hox-free mESCs and mandibular CNCCs finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| H3K27ac ChIP-seq | E10.5 mouse CNCCs (FNP, maxillary, mandibular, PA2) | none | super-enhancer calling via merged H3K27ac peaks (ROSE-based workflow) | — |
| Promoter Capture Hi-C (PCHi-C) | E10.5 FNP/Mx/Md/PA2 CNCCs and E12.5/E14.5 pinna CNCCs | none | significant promoter-enhancer interactions (CHiCAGO score) | biotinylated RNA bait probes |
| Hi-C | mouse embryonic stem cells (mESCs), E10.5 Md/PA2 CNCCs, E12.5/E14.5 pinna CNCCs | none | 3D chromatin interactions, TAD boundaries/separation score | HiCExplorer hicFindTADs |
| RNA-seq | E10.5 FNP/Mx/Md/PA2 CNCCs and E12.5/E14.5 pinna CNCCs | none | gene expression levels (RPKM, logFC) | — |
| ATAC-seq | E10.5 CNCC subpopulations and E12.5/E14.5 pinna CNCCs | none | chromatin accessibility at Hoxa locus and HIRE1/HIRE2 | — |
| ChIP-seq (H3K27me3, H3K4me2) | E10.5 Md/PA2 CNCCs and E12.5/E14.5 pinna CNCCs | none | repressive/active histone mark enrichment across Hoxa cluster | — |
| CRISPR-mediated genomic deletion | mouse (in vivo, CNCC-derived craniofacial skeleton) | targeted deletion of HIRE1 or HIRE2 (also on Hoxa2 haploinsufficient background) | craniofacial/skeletal phenotype and Hoxa2/Hoxa3 transcriptional levels | — |
| ChIP-seq (Hoxa2, Pbx, Meis) | PA2 CNCCs at E11.5 | none | transcription factor binding sites at HIRE1/HIRE2 and Hoxa2 locus | — |
- – 2232 putative SEs identified as active in at least one of four CNCC subpopulations
- – 147 SEs linked to 62 positional transcription factor genes (148 unique SE-promoter pairings), including Hoxa2, Msx1/2, Tfap2b, Pax3, Alx4, Six1/2, Alx1, Pitx1, Barx1, Meis1/2, Dlx3, Hand2
- – 5 SEs (SE1-5), subdivided into HIRE1 (175 kb, ~1.07 Mb from Hoxa2) and HIRE2 (39 kb, ~1.33 Mb from Hoxa2), selectively interact with Hoxa2 in PA2 CNCCs but not in Md/Mx/FNP CNCCs or mESCs
- – Hoxa2 expression increased from E10.5 to E12.5 pinna CNCCs, then decreased from E12.5 to E14.5 logFC=0.753 (up, FDR=8.89E-07); logFC=-0.455 (down, FDR=5.71E-05)
- ▼ HIRE1 deletion reproduces the full homeotic Hoxa2 knockout phenotype and causes additional PA3/PA4 skeletal abnormalities with downregulated Hoxa2 and Hoxa3 transcription
- ▼ HIRE2 deletion alone shows no major skeletal changes, but combined with Hoxa2 haploinsufficiency produces microtic (small, malformed) pinnae
- – HIRE1 and HIRE2 conserved elements found down to birds (and partly fish/coelacanth) across 60 vertebrates
- – Inter-TAD Hoxa2-HIRE interactions appear as an architectural stripe on Hi-C, present only in Hoxa2-expressing PA2/pinna CNCCs, absent in Hox-negative mESCs and Md CNCCs
- count 2232 (total putative super-enhancers identified genome-wide across CNCC subpopulations)
- count 147 SEs linked to 62 genes; 148 unique pairings (SEs linked to positional transcription factor-coding gene promoters)
- fold_change logFC = 0.753, FDR = 8.89E-07 (Hoxa2 expression increase from E10.5 PA2 to E12.5 pinna CNCCs)
- fold_change logFC = -0.455, FDR = 5.71E-05 (Hoxa2 expression decrease from E12.5 to E14.5 pinna CNCCs)
- other HIRE1 = 175 kb region, ~1.07 Mb from Hoxa2 (size and distance of HIRE1 regulatory subdomain (SE2-4))
- other HIRE2 = 39 kb region, ~1.33 Mb from Hoxa2 (size and distance of HIRE2 regulatory subdomain (SE5))
- other CHiCAGO score ≥5 (significant threshold), max color scale at ≥20 (PCHi-C interaction significance and visualization threshold)
- count 22 elements in HIRE1 conserved to birds (2 also in fish); 6 elements in HIRE2 conserved to birds (2 in coelacanth) (cross-species basewise sequence conservation analysis across 60 vertebrates)
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 genomics/epigenomics study characterizes super-enhancers (SEs) in mouse cranial neural crest cell (CNCC) subpopulations using H3K27ac ChIP-seq, ATAC-seq, RNA-seq, Hi-C, and Promoter-Capture Hi-C (PCHi-C). SEs were identified genome-wide via the ROSE algorithm from merged H3K27ac peaks, and SE–promoter interactions were assessed using CHiCAGO scores from PCHi-C data generated in biological duplicate across four CNCC subpopulations and three developmental time points. Differential gene expression across time points was quantified as log fold-change with FDR-adjusted p-values, and TAD boundaries were called computationally with HiCExplorer's hicFindTADs. CRISPR-mediated deletions of HIRE1 and HIRE2 were evaluated phenotypically in wild-type and Hoxa2 haploinsufficient backgrounds.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| CHiCAGO statistical framework for promoter-capture Hi-C interaction scoring (score threshold ≥5) | SE–promoter interaction calling genome-wide across all four CNCC subpopulations (FNP, Mx, Md, PA2) at E10.5 and pinna CNCCs at E12.5 and E14.5 | — | not stated |
| ROSE algorithm signal-rank threshold for super-enhancer identification (geometric inflection point on ranked H3K27ac signal; not a classical hypothesis test) | Genome-wide SE calling from merged H3K27ac ChIP-seq peaks across all four CNCC subpopulations | — | not stated |
| RNA-seq differential expression test (method not named in provided text; results reported as logFC with FDR) | Hoxa2 expression change E10.5 PA2 vs. E12.5 pinna (logFC=0.753, FDR=8.89E-07) and E12.5 vs. E14.5 pinna (logFC=−0.455, FDR=5.71E-05) | Duplicate libraries per time point; exact number of pooled embryos not stated in provided text | not stated |
| HiCExplorer hicFindTADs TAD separation score (computational boundary-calling method; not a classical hypothesis test) | TAD boundary identification in Hi-C data at 25 kb resolution across Md, PA2, E12.5, and E14.5 pinna CNCCs | — | not stated |
| k-means clustering on RNA-seq gene expression levels | Heatmap grouping of 147 SE–promoter pairs linked to positional transcription factor-coding genes (Fig. 1b; 148 unique pairings, 62 genes) | 147 SEs linked to 62 genes | not stated |
-
Super-enhancers were identified using the ROSE algorithm, which merges H3K27ac peaks within 12.5 kb and applies a geometric inflection-point threshold on ranked signal↳ Could also: Alternative SE calling strategies such as HOMER annotation-based ranking, ChromHMM/Segway hidden Markov model chromatin state segmentation, or a permutation-derived signal threshold could also be applied — Different algorithms use different merging distances and thresholds, so reporting the sensitivity of the 2232 SE calls to algorithmic choices would allow readers to gauge how boundary-dependent the downstream biological conclusions are
-
PCHi-C interaction significance was assessed using a fixed CHiCAGO score threshold of ≥5↳ Could also: FitHiC2, HOMER, or a data-adaptive threshold selected by permutation against distance-matched random loci could also be used to call significant interactions — Reporting how the key inter-TAD SE–Hoxa2 interactions behave across a range of score thresholds (e.g., ≥4, ≥5, ≥6) would clarify how robustly those contacts are detected above background
-
PCHi-C libraries were generated in biological duplicate (n=2) for each CNCC subpopulation↳ Could also: Three or more biological replicates would enable formal statistical modeling of replicate variability using DESeq2-based count models (as implemented in diffHic or multiHiCcompare) for interaction-level differential testing between cell types — With n=2, formal inter-group differential interaction testing is underpowered; additional replicates would allow quantitative comparison of interaction frequencies across subpopulations and time points rather than relying on visual inspection of CHiCAGO scores
-
The RNA-seq differential expression method was not named in the provided text, though results were reported as logFC with FDR↳ Could also: Explicitly naming and citing the statistical model (e.g., DESeq2 negative-binomial Wald test or edgeR quasi-likelihood F-test) is standard practice — Different count-based DE methods make different assumptions about overdispersion; naming the method aids reproducibility and lets readers evaluate whether the chosen model is appropriate given the low replicate count
-
Differential expression effect sizes were reported as logFC without confidence intervals↳ Could also: Reporting 95% confidence intervals on logFC alongside FDR-adjusted p-values is also standard and conveys estimation precision — Confidence intervals communicate uncertainty in the effect estimate separately from the significance decision, which is especially informative when replication is limited and the FDR value alone does not reflect estimation variance
-
Heatmap rows were grouped by k-means clustering on gene expression levels across CNCC subpopulations↳ Could also: Hierarchical clustering with Ward linkage or model-based clustering (e.g., Mclust) could also organize SE–promoter pairs, and bootstrap stability assessment would complement any clustering method — k-means requires specifying k in advance and results can vary with random initialization; hierarchical methods yield a deterministic dendrogram that may reveal nested expression patterns, and reporting cluster stability would strengthen the grouping interpretation
Result convergence & founder nodes
Findings this paper shares with others that ran a comparable experiment. A node’s strength is how many independent papers report it (replication breadth) — not how often it is cited, so a heavily-replicated but under-cited founder still stands out.
-
HIRE1 and HIRE2 are Hoxa2 super-enhancer clusters spanning 175 kb and 39 kb and located approximately 1.07 Mb and 1.33 Mb 3' of the Hoxa2 promoter, residing in an adjacent TADChIP-seq mouse cranial-neural-crest 2023×1papers★ This paper is the founder (earliest)
-
Five H3K27ac super-enhancers located more than 1 Mb 3' of Hoxa2 form significant promoter-capture contacts with Hoxa2 selectively in pharyngeal arch 2 cranial neural crest cells, representing inter-TAD regulatory interactionsHi-C mouse cranial-neural-crest 2023×1papers★ This paper is the founder (earliest)
-
HIRE1 super-enhancer deletion phenocopies full Hoxa2 knockout with homeotic transformation of PA3/PA4 skeletal derivatives and reduces Hoxa2 and Hoxa3 transcription in cranial neural crest cellsother mouse cranial-neural-crest down 2023×1papers★ This paper is the founder (earliest)
-
HIRE2 super-enhancer deletion on a Hoxa2 haploinsufficient background produces microtic pinnae, demonstrating HIRE2 contributes to Hoxa2 dosage in pinna morphogenesisother mouse pinna down 2023×1papers★ This paper is the founder (earliest)
-
Hoxa2 expression rises from E10.5 PA2 to E12.5 pinna cranial neural crest cells then declines by E14.5, indicating stage-specific transcriptional dynamicsRNA-seq mouse cranial-neural-crest mixed 2023×1papers★ This paper is the founder (earliest)
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-37277355
Paper: Kessler et al. 2023, Nat Commun 14:3434. "A multiple super-enhancer region establishes inter-TAD interactions and controls Hoxa function in cranial neural crest." DOI 10.1038/s41467-023-38953-0.
Named code artifact (RU): https://github.com/4dn-dcic/hic2cool (v0.8.3) — a third-party Hi-C format converter. Per BRIEF rule P16, applying this existing tool to the paper's own data is a fully valid reproduction.
Real data accessions (corrected at run time)
The scaffold's data.json listed GSE89437, but that is the group's prior
2017 SuperSeries (PMID 28360266; ChIP/ATAC/RNA only, no Hi-C) which this paper
re-uses. The 2023 paper's own data (from its Data Availability statement) is:
| Accession | Assay | Note |
|---|---|---|
| GSE211904 | SuperSeries | all data |
| GSE211899 | ATAC-seq | |
| GSE211900 | ChIP-seq | H3K27ac → super-enhancers |
| GSE211901 | Hi-C | in scope — ships processed .hic |
| GSE211902 | PCHi-C | |
| GSE211903 | RNA-seq |
The Hi-C pipeline (verbatim from Methods)
"The Hi-C data has been mapped and quality controlled with HiCUP (version 0.6.1), the interaction matrix was created with Juicer (version 1.6), and was transformed into a cool matrix with hic2cool (https://github.com/4dn-dcic/hic2cool, version 0.8.3) and HiCExplorer's hicConvertFormat (version 3.7.2). ... The merged cool matrix was normalized with HiCExplorer's hicCorrectMatrix (version 3.7.2) using KR normalization. The TAD calling was applied by HiCExplorer's hicFindTADs (version 3.7.2). The Hi-C and TAD data was plotted with pyGenomeTracks (version 3.7)."
Genome: mm10 / GRCm38. Hi-C heatmaps shown at 25 kb resolution (Fig. 5b).
IN SCOPE (pipeline-derived, well-specified, low-hanging)
The paper ships the Juicer .hic matrices directly in GSE211901 (e.g.
GSM6505199_PA2_E10.5wt_HiC_1stBioRep.hic, 393 MB). Starting from these shipped
processed matrices, we reproduce the documented downstream pipeline exactly with
the named tools:
- hic2cool 0.8.3 — convert
.hic→.cool/.mcool(the RU's named tool). Deterministic; verify bins/resolutions/contact totals. - HiCExplorer 3.7.2 —
hicCorrectMatrixKR-normalize the 25 kb matrix;hicFindTADsto call TAD separation scores / boundaries. - Compare the called TAD structure around the Hoxa locus (chr6 ~50.4–52.6 Mb)
to the paper's reported features (Fig. 5b):
- domain encompassing the Hoxa cluster + Evx1 at chr6:52,145,433–52,327,518;
- a TAD boundary between Hoxa2 and the HIRE1/HIRE2 super-enhancer region.
Pipeline per result: hic2cool (convert) + HiCExplorer hicCorrectMatrix/hicFindTADs (normalize + TAD call), on the paper's shipped PA2 E10.5 Hi-C matrix.
OUT OF SCOPE (not attempted — and why)
- Raw-read mapping (HiCUP 0.6.1 → Juicer 1.6): the paper ships the resulting
.hicmatrices, so we start from those (80/20: re-mapping raw FASTQ is the hard, redundant last 20%; it would re-derive a file the authors already provide). - Super-enhancer / HIRE calling (ROSE on H3K27ac ChIP-seq): different assay (GSE211900), different pipeline; not the named RU tool.
- PCHi-C CHiCAGO interaction calling, virtual-4C: GSE211902; separate pipeline.
- All wet-lab / mouse-genetics phenotypes (microtia, homeotic transformation): not computational.
No completeness claim
We reproduce one clearly-specified computational sub-pipeline (hic2cool conversion + HiCExplorer TAD calling on the PA2 Hi-C matrix) and compare the TAD structure around Hoxa to the paper's reported coordinates. We do not reproduce the full figure set.
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.
This is a clean, faithful reproduction of the Fig.5b Hi-C/TAD claims. Starting from the authors' own shipped Juicer .hic (GSM6505199) and running the documented hic2cool 0.8.3 -> HiCExplorer 3.7.2 KR -> hicFindTADs pipeline at 25 kb, the highlighted Hoxa+Evx1 domain (reported chr6:52,145,433-52,327,518) reproduced as chr6:52,150,000-52,325,000 with both edges inside a single 25 kb bin, plus the reported Hoxa2/HIRE boundary near ~52.15 Mb. The only deviations are sub-bin coordinate rounding, and the result is independently corroborated by the Md E10.5 sample. The single caveat is on our scaffold's side (accession GSE89437 was a wrong, Hi-C-free SuperSeries, corrected by the agent to GSE211901) — not an authors' or derivability problem.
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.