Nucleosome regulatory dynamics in response to TGFβ.
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
The paper's core computational pipeline (SOLiD colorspace MNase-seq alignment -> sictin signal binarization -> SuMMIt Bayesian modeller -> SuMMIt change differential-occupancy test) was reproduced end-to-end on chromosome 1 for the two ArrayExpress runs that actually have deposited sequence data (ctrl_rep1, treat_rep3 of the 8 declared in E-MTAB-1750). Two independent, non-trivial engineering obstacles had to be solved, both symptomatic of running a 2013-era pipeline against a modern software environment rather than any flaw in the paper: (1) the SOLiD colorspace reads required bowtie1 with -C against a colorspace-encoded index, not bowtie2 (fixed pre-session); (2) the repo-shipped modeller/change binaries were linked against GSL 1.x (libgsl.so.0), unavailable in an environment that only ships GSL 2.8, and simply rebuilding from source initially failed too because the Makefile's hardcoded -I/usr/include header path collided with the conda-toolchain's own bundled glibc headers -- removing that hardcoded path (GSL 2.8 headers already reachable via CPATH) let the rebuild succeed and the binaries linked cleanly against libgsl.so.28. Once fixed, modeller converged on biologically sane parameters (estimated nucleosome footprint length 147bp in both conditions, matching the paper's mononucleosome design) and change produced full genome-position odds-ratio tracks for chr1. This is graded 'partial' overall, not 'reproduced', because: only 2 of the 8 declared sequencing runs in E-MTAB-1750 have actual deposited data (a dataset-completeness problem, not a pipeline problem), the reproduction was deliberately scoped to chr1 only rather than genome-wide, and the paper does not publish a machine-readable numeric result to grade the SuMMIt output against directly -- so the claims are graded on 'does the pipeline run and produce internally-consistent, plausible output' rather than 'exact number match'. Not attempted: any wet-lab/microscopy results in the paper (out of scope per the brief), genome-wide (all-chromosome) processing, and the 6 missing sequencing runs (cannot be reproduced -- no data exists for them in the archive).
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.
✎ 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
- 2026-08-01
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-08-01no human curator yet
- Last updated
- 2026-08-01
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: opusCan the human nucleosomal landscape and its dynamics be precisely mapped in response to an external stimulus, and does TGFβ1-induced nucleosome depletion relate to transcription factor binding and gene expression changes? The authors test this in HepG2 hepatic cells before and after TGFβ1 stimulation using a new Bayesian positioning method (SuMMIt).
- ★ SuMMIt, a Bayesian strand-based mixture model requiring support from both ends of sequenced fragments, enables precise nucleosome mid-position calling, fuzziness scoring and between-condition change detection. method
- ★ The canonical average nucleosome pattern around TSSs and flanking TF binding sites is present at very few individual loci; the double peak surrounding TF binding sites is an artifact of averaging over many loci. finding
- ★ TGFβ1 stimulation for 1 h depletes nucleosomes at 24 318 loci in HepG2 cells relative to unstimulated cells. finding
- ★ Loci with TGFβ1-induced nucleosome depletion are enriched for transcription factor binding motifs, with 44–78% over-represented depending on genomic annotation category. finding
- ★ Nucleosome depletion at these loci is accompanied by altered TF binding, verified for HNF4α by ChIP-qPCR. mechanism
- ★ Many loci with depleted nucleosomes are associated with gene expression changes measured by RNA sequencing. finding
- Nucleosome phasing varies by genomic context: phased, intermediate and fuzzy nucleosomes are distributed differently across exonic, intronic and intergenic regions and depend on exon length. finding
- A genome-wide nucleosome map of HepG2 cells before and after TGFβ1 stimulation, plus the platform-independent C++ SuMMIt implementation, is provided as a resource. resource
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| MNase digestion + mononucleosome sequencing (MNase-seq) | Human hepatocellular carcinoma HepG2 cells | TGFβ1 2.5 ng/ml for 1 h vs untreated control (serum-starved overnight in 1% FBS) | Nucleosome occupancy / inferred nucleosome mid-positions, interior regions, fuzziness and depletion loci | SOLiD fragment library protocol, SOLiD v3 chemistry, one slide per emulsion PCR; Qiagen gel extraction kit; 147-bp mononucleosome fragments excised from 2% agarose |
| Strand-specific RNA sequencing (RNA-seq) | HepG2 cells | TGFβ1 stimulation vs unstimulated control | Transcript coverage and RPKM counts per gene; fold change between TGFβ1 and control | SOLiD whole transcriptome library kit (Ambion), RiboMinus rRNA depletion (Invitrogen), AmpliTaq 18 cycles, SOLiD v3 chemistry, two slides per library; BioAnalyzer (Agilent) for RNA integrity; TRIzol-chloroform (Invitrogen) RNA prep |
| TaqMan real-time quantitative RT-PCR (RNA-seq validation) | HepG2 cells | TGFβ1 stimulation vs unstimulated control | Relative expression by comparative Ct method with GAPDH as reference; log2 fold change for 25 selected genes | Applied Biosystems 7000 Real-Time PCR System with SDS software 1.2.3; TaqMan Gene Expression Master Mix and Assay Mix (Applied Biosystems); iScript cDNA Synthesis Kit (Bio-Rad); DNase I (Qiagen) |
| Chromatin immunoprecipitation followed by qPCR (ChIP-qPCR) | HepG2 cells | TGFβ1 stimulation vs unstimulated control | HNF4α binding fold enrichment between TGFβ1-stimulated and control cells at 10 candidate sites with predicted nucleosomal relocation, normalized to 3 negative sites | anti-HNF4α antibody SC-6556; SYBR green qPCR, triplicates |
| Computational nucleosome positioning and differential analysis (SuMMIt) | HepG2 nucleosome sequencing data (unstimulated and TGFβ1-stimulated), trained separately | none (in silico) | Per-strand log-odds of nucleosome mid-position vs background, change log-odds between samples, fuzziness (positional standard deviation) | C++ using GNU Scientific Library; Gibbs sampling; SICTIN software suite for footprints; BEDTools for locus partitioning; ENSEMBL release 54 (NCBI 36) annotations |
| TF motif over-representation analysis | 130-bp sequences centered on nucleosome interior regions in TGFβ− cells associated with depletion in TGFβ+ cells, per annotation category | none (in silico); background = twice as many random non-depleted nucleosome loci from the same category | Over-represented TF motifs (raw score > 6, P < 0.05) | Clover; JASPAR database position weight matrices, clade vertebrates |
| Meta-profile / heat map and K-means clustering of nucleosome signal | HepG2 nucleosome data at TSSs of top 5000 highly expressed protein-coding genes and at 25 651 JUND binding sites; compared with DNaseI hypersensitivity | TGFβ1 stimulated vs unstimulated | Average log-odds footprints, counts of fragment-length (147 bp) extended reads, and 10 K-means clusters from discretized log-odds in 2-kb windows | SICTIN software suite; UCSC Genome Browser for track visualization |
- ▼ Systematic search identified loci with nucleosomes present in unstimulated cells but depleted in TGFβ1-stimulated cells 24 318 loci
- ▲ Loci with depleted nucleosomes were over-represented for TF binding motifs, varying by genomic annotation category 44–78%
- – HNF4α binding changes at regions with ejected nucleosomes were verified by ChIP-qPCR in TGFβ1-stimulated versus control cells
- – Many loci with nucleosome depletion were associated with expression changes measured by RNA-seq
- – The average nucleosome pattern (including the double peak flanking TF binding sites) was present at very few individual loci, as revealed by K-means clustering of TSS and JUND site profiles
- – TaqMan qRT-PCR validated RNA-seq fold changes for selected genes (e.g. SMAD7 induction shown by RNA-seq coverage)
- – Nucleosomes distributed across exonic, intronic and intergenic regions differently from the genomic sequence coverage of these regions
- – The fraction of phased, intermediate and fuzzy nucleosomes in exons varied with exon length
- count 24 318 (Loci with nucleosome depletion in TGFβ1-stimulated versus unstimulated HepG2 cells)
- other 44–78% (Fraction of depleted-nucleosome loci over-represented for TF binding motifs, depending on genomic annotation)
- count 25 651 (JUND (JunD) binding sites used for average footprints and heat map clustering)
- count 5000 (Top highly expressed protein-coding genes whose TSSs were profiled)
- other 2.5 ng/ml TGFβ1, 1 h treatment (Stimulation dose and duration for HepG2 cells)
- other log-odds of change > 10 (Filter threshold for calling nucleosome depletion between samples)
- other fuzziness ≤60 phased, 60–80 intermediate, >80 fuzzy (Thresholds for nucleosome positional concordance classes)
- count 10 candidate sites; 3 negative sites; triplicate qPCR (HNF4α ChIP-qPCR validation design)
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.
The study profiles genome-wide nucleosome occupancy and RNA expression in HepG2 cells before and after 1 h of TGFβ1 stimulation using a custom Bayesian method (SuMMIt) that fits Gibbs-sampled Poisson mixture models to call nucleosome mid-positions and identify their loss between conditions via a fixed log-odds change threshold. Over-representation of transcription-factor binding motifs at nucleosome-depleted loci was assessed with the Clover program using a raw-score and P-value (<0.05) cutoff against matched background sequences. Predicted changes in HNF4α binding and RNA-seq-derived expression changes were checked against independent triplicate qPCR (fold-enrichment) and TaqMan comparative Ct assays, compared descriptively/graphically to the sequencing-based estimates. Overall, results are reported largely as thresholded scores, log-odds/log2 fold-changes, and descriptive comparisons rather than through classical inferential hypothesis tests with significance values computed across biological replicates.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Clover motif over-representation test (raw score threshold with P < 0.05) | identification of over-represented transcription factor binding motifs in 130-bp sequences centered on nucleosome-depleted loci, compared to matched background loci, separately by genomic annotation category | 24 318 nucleosome-depletion loci vs. twice as many randomly selected background loci per annotation category | not stated |
| Log-odds change score from Bayesian mixture model (SuMMIt), fixed threshold calling | identifying nucleosome mid-positions present in unstimulated cells and depleted in TGFβ-stimulated cells | genome-wide comparison of one sequenced sample per condition; threshold of log-odds of change > 10 | not stated |
| Fold-enrichment comparison (ChIP-qPCR) | HNF4α ChIP-qPCR at 10 candidate binding sites with predicted nucleosome relocation, normalized to 3 negative control sites | triplicate qPCR reactions per site | na |
| Comparative Ct method (TaqMan qPCR) vs. RNA-seq log2 fold-change | validation of RNA-seq-derived expression changes for 25 selected genes (Figure 8C) | triplicate 25-µl TaqMan reactions per gene | na |
-
TF motif enrichment was assessed with Clover using a raw-score and P < 0.05 threshold across many motifs simultaneously, without a stated multiple-testing correction.↳ Could also: Apply a multiple-testing correction (e.g., Benjamini-Hochberg FDR) across the family of tested motifs. — When many motifs are tested at once, an FDR correction would control the expected proportion of false positives among motifs called enriched, complementing the raw-score/P-value cutoff already used.
-
Differential nucleosome occupancy between TGFβ-stimulated and unstimulated cells was called from a single sequenced sample per condition using a fixed Bayesian log-odds threshold.↳ Could also: Sequence independent biological replicates per condition and apply a count-based differential test (e.g., a DESeq2- or edgeR-style generalized linear model) to occupancy counts. — Replicate-based testing would let biological variability be estimated directly and would yield formal p-values/FDR for called changes, complementing the current single-sample threshold approach.
-
RNA-seq expression changes were summarized as RPKM-based log2 fold-change from a single TGFβ+ vs. TGFβ- comparison, without a stated statistical significance test.↳ Could also: Use a dedicated RNA-seq differential expression tool (e.g., DESeq2 or edgeR) with biological replicates. — This would provide moderated fold-change estimates with adjusted p-values, letting readers gauge which expression changes are statistically supported versus a purely descriptive fold-change ranking.
-
qPCR validation of HNF4α ChIP enrichment compared TGFβ+ vs. TGFβ- fold enrichment using triplicate technical reactions, without a stated formal significance test or biological-replicate variability measure.↳ Could also: Report SD/SEM across independent biological replicates and apply a paired t-test (or similar) to the fold-enrichment values. — This would quantify whether the observed binding differences exceed expected experimental variability, beyond a single fold-enrichment estimate.
-
Nucleosome fuzziness (positional spread) is reported as the standard deviation of read positions around inferred mid-positions.↳ Could also: Report this dispersion alongside a 95% CI or IQR, particularly if the underlying distribution of read positions is skewed. — SD assumes a roughly symmetric spread; an IQR or CI can convey uncertainty more robustly when distributions are skewed or heavy-tailed.
-
TaqMan qPCR validation of RNA-seq fold-changes for 25 genes was presented as paired log2 fold-change values without a stated correlation statistic.↳ Could also: Report a correlation coefficient (e.g., Pearson or Spearman r) with a confidence interval between RNA-seq and TaqMan fold-changes. — A quantitative correlation statistic would formally summarize agreement between the two platforms, complementing the visual/graphical comparison already shown.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
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 paper's full computational chain (bowtie1 -C colorspace alignment → sictin build_binaries → SuMMIt modeller → SuMMIt change) was successfully re-executed on chr1 and produced internally consistent output: a 147 bp mononucleosome model in both conditions and condition-distinct mixture parameters (pFpeak 0.145 ctrl vs 0.058 treat; lambdaFpeak 140.9 vs 332.0), plus complete chr1 odds-ratio tracks. The binding limitation is on the authors'/archive side, not analytical: E-MTAB-1750 declares 8 runs but only 2 have retrievable sequence data, and the paper publishes no machine-readable numeric result, so there is literally no reported value to diff against. Additional friction was pure legacy-toolchain drift (libgsl.so.0 missing; Makefile -I/usr/include collision) and our own chr1-only scoping — neither indicates an error by the authors. Verdict: a credible partial reproduction with no sign of fabrication or miscomputation, but the central genome-wide TGFβ nucleosome-dynamics claim remains unverifiable from the deposited material.
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.