Corpus 1,272 assessed · 1,173 scored · 643 reproduced ≥75 · 168 flagged ·∅ 74.1/100
← New search

SOX10-regulated promoter use defines isoform-specific gene expression in Schwann cells.

BMC Genomics · 2020
L1 94/100 3/4
Why this verdict

The main results reproduced: recomputed values matched the published ones within tolerance.

Reproduced on the brainbox compute brainarbeit.com
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +1
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • 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
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
94/100
Reproducibility score
1.1 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 87% of all assessed papers rank 133 of 1173 scored

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 is a Tn5Prime (CAGE-like) TSS/promoter-usage study in rat Schwann cells; the pipeline-derived results are the candidate-TSS classification (Fig 1a) and edgeR differential-expression counts (Fig 2a cAMP-vs-control, Fig 3a delSOX10-vs-parental). All 9 specific reported count statistics reproduce EXACTLY from the deposited GSE139321 processed table (4993 candidate TSSs; 2993 loci; 4044 expressed / 465 up / 401 down; 265 down / 100 up / 3766 unchanged / 862 not-expressed). NOTE these are recounts of the authors' OWN deposited edgeR log2FC/FDR (RPM-only deposit -> edgeR not re-runnable from scratch); they confirm figures match the deposit (no fabrication) but are not an independent pipeline re-execution. The one fully INDEPENDENT computation - a BEDTools 1-kb intersection of the 4993 TSSs vs the two cited public ChIP-seq peak files - reproduces H3K4me3 exactly (4993/4993) and positively identifies the SOX10 file (GSE64703 PNS, 99.3% of assigned peaks exact members); the 17% per-TSS-window shortfall is fully explained as locus-level (alternative-promoter) peak assignment, consistent with the paper's premise. NOT attempted: the full 39,706-TSS denominator (full TSS universe not deposited; needs STAR/Paraclu on raw FASTQ), wet-lab luciferase/Western values, similaRpeak/metagene AUC permutation p-values, motif/PhastCons/Tau distributions, and manual human-RefSeq isoform curation (out of scope or only partly pipeline-derived). Code link in brief (adeschen/similaRpeak) is a genuinely-cited third-party tool but only for a secondary result; core results need only BEDTools + edgeR on public processed files.

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.

  1. v1 current initial assessment Score 94
    assessed: 2026-06-18 ⛓ a9413159a06e
✎ 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.

Reason for the rerun

We email you a confirmation link first. The rerun is an objective re-measurement — it cannot change the verdict in your favour, only ask us to look again.

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
2026-06-18
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-18
no 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: opus
Founding hypothesis

Because SOX10 is essential for Schwann cell myelination and regulates alternative promoters that direct isoform-specific gene expression, the authors sought to define SOX10-mediated promoter and transcription start site (TSS) use genome-wide in Schwann cells to prioritize candidate target genes and isoforms for further study.

Core claims
  • SOX10 binds preferentially at proximal promoter elements directly over TSSs in Schwann cells in vivo, identifying candidate SOX10-regulated promoters. finding
  • 4993 sciatic nerve TSSs (associated with both H3K4me3 and SOX10 ChIP-Seq peaks) represent high-confidence candidate SOX10-regulated promoters mapping to 2993 loci. resource
  • A substantial fraction of candidate SOX10-regulated loci harbor multiple TSSs and distinct protein-coding isoforms, implicating SOX10 in isoform-specific gene expression. finding
  • TSS usage at SOX10-associated promoters changes during cAMP-induced primary Schwann cell differentiation, with up/down-regulated groups enriched for distinct functional categories. finding
  • Genome-wide TSS profiling (Tn5Prime) combined with mining of SOX10/H3K4me3 ChIP-Seq datasets defines SOX10-regulated promoter activity across three Schwann cell models. method
  • ARPC1A, CHN2, DDR1, and GAS7 are validated as previously unreported SOX10 target loci, suggesting roles in PNS myelination. finding
  • Upregulated TSSs upon differentiation are enriched for cholesterol/sterol biosynthesis, protein targeting to ER/membrane, translation, and myelination functions. finding
  • Downregulated TSSs upon differentiation are enriched for intracellular vesicle sorting/transport, cortical cytoskeleton regulation, and the unfolded protein response. finding
Experimental setups
Assay System Perturbation Readout Platform
Tn5Prime TSS sequencing (5' template-switching, global TSS mapping) adult rat sciatic nerve (age 6–9 months) none transcription start site identity and transcript abundance Tn5Prime library preparation
Tn5Prime TSS sequencing rat primary Schwann cells (differentiation paradigm) CPT-cAMP treatment (drug-induced differentiation) differential TSS usage/expression between untreated and cAMP-treated cells Tn5Prime library preparation
ChIP-Seq dataset mining/intersection (SOX10 and H3K4me3) rat sciatic nerve none SOX10 and active-promoter (H3K4me3) peak overlap with TSSs; metagene aggregate SOX10 signal
Gene ontology enrichment analysis differentiating primary Schwann cell TSS gene sets na enriched GO terms with fold enrichment and FDR p-values
Functional promoter validation studies cultured Schwann cells with ablated SOX10 function SOX10 loss/ablation validation of SOX10-regulated promoter activity at ARPC1A, CHN2, DDR1, GAS7
Key results
  • 4993 of 39,706 sciatic nerve TSSs (12.6%) reside within 1 kb of an H3K4me3 peak overlapping a SOX10 ChIP-Seq peak 12.6% (4993/39,706)
  • These 4993 TSSs map to 2993 unique loci including known SOX10 targets (Mpz, Mbp, Pmp22) 2993 loci
  • 7455 TSSs (18.8%) map to H3K4me3 but not SOX10; 431 (1.1%) map to SOX10 but not H3K4me3; 26,827 (67.6%) map to neither 18.8% / 1.1% / 67.6%
  • 739 of 2993 loci (25%) have ≥2 RefSeq transcripts from unique TSSs; 525 (17.5%) include multiple TSSs encoding distinct protein-coding sequences 25% and 17.5%
  • 4044 of 4993 candidate TSSs (81%) expressed in control and/or cAMP-treated primary Schwann cells 81% (4044/4993)
  • 465 TSSs (11.5%) upregulated and 401 (9.9%) downregulated with CPT-cAMP treatment 11.5% up, 9.9% down
  • Candidate SOX10-bound loci enriched for 'regulation of myelination' (GO:0031641) FDR p=0.0324
  • SOX10 ChIP-Seq signal concentrated directly over TSSs near H3K4me3 peaks, with stronger signal at dual-peak TSSs
Key statistics
  • count 4993 of 39,706 TSSs (12.6%) (TSSs within 1 kb of overlapping H3K4me3 and SOX10 ChIP-Seq peaks)
  • pvalue FDR-corrected p = 0.0324 (enrichment for 'regulation of myelination' (GO:0031641) among 2993 candidate loci)
  • count 465 upregulated (11.5%), 401 downregulated (9.9%) (differentially expressed TSSs upon CPT-cAMP treatment of primary Schwann cells)
  • fold_change 17.49 fold enrichment (cholesterol biosynthetic process (GO:0006695), FDR p=1.55E-08, upregulated TSSs)
  • fold_change 17.04 fold enrichment (SRP-dependent cotranslational protein targeting to membrane (GO:0006614), FDR p=2.45E-19)
  • fold_change 10.59 fold enrichment (myelination (GO:0042552), FDR p=4.51E-10, upregulated TSSs)
  • fold_change 33.96 fold enrichment (vesicle transport along actin filament (GO:0030050), FDR p=2.44E-02, downregulated TSSs)
  • count 739 of 2993 (25%); 525 of 2993 (17.5%) (loci with multiple unique-TSS transcripts / multiple TSSs encoding distinct proteins)

Statistical methods review

Model: sonnet

A 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 combined Tn5Prime TSS-sequencing with published SOX10 and H3K4me3 ChIP-seq data to classify active TSSs in adult rat sciatic nerve and primary Schwann cells as candidate SOX10-regulated promoters using proximity-based overlap. Differential TSS expression between untreated and cAMP-differentiated primary Schwann cells was assessed with FDR-corrected p-values (threshold 0.05), and gene ontology enrichment with FDR correction was performed on differentially expressed TSS groups. Statistical comparisons of expression distributions and aggregate ChIP-seq signals across TSS classes were reported with asterisk-coded p-value thresholds in figures; underlying test names and software were not specified.

Replicationunclear Sample sizeNumber of biological replicates for sciatic nerve and primary Schwann cell experiments not stated in the main text GroupsFour TSS classes by ChIP-seq overlap (SOX10+H3K4me3, H3K4me3-only, SOX10-only, neither); untreated vs. cAMP-differentiated primary Schwann cells Pairingunclear Randomization/blindingnot stated Dispersionnone Exact p-valuesno Effect sizesno Confidence intervalsno Multiplicity correctionFDR correction; specific algorithm (e.g., Benjamini-Hochberg) not stated
Statistical tests used
Test Applied to n Assumptions
Differential expression test (specific method not stated); FDR-corrected p < 0.05 TSS expression changes between untreated and cAMP-treated primary Schwann cells (Fig. 2a, Supplementary Table 1) not stated
Gene ontology enrichment analysis with FDR correction (specific algorithm not stated) Functional enrichment of upregulated (465 TSSs / 355 loci) and downregulated (401 TSSs / 372 loci) TSS groups upon cAMP treatment (Tables 2–3) 465 upregulated TSSs; 401 downregulated TSSs not stated
Statistical test comparing cumulative TSS expression distributions across classes (specific test not stated); reported as p < 0.00001 Cumulative expression-level distributions of TSS classes in sciatic nerve (Fig. 1b) not stated
Statistical test comparing aggregate SOX10 ChIP-seq signal across TSS classes (specific test not stated); reported as p < 0.01 Metagene aggregate SOX10 ChIP-seq signal in 2-kb window around TSS classes (Fig. 1c) not stated
Statistical test comparing in-vivo expression distributions across TSS classes in primary Schwann cells (specific test not stated); reported as p < 0.03 Cumulative in-vivo expression-level distributions of TSS classes in primary Schwann cells (Fig. 2b) not stated
Approaches that could also have been used
  • Differential TSS expression was assessed using FDR-corrected p-values without naming the underlying statistical model or software
    Could also: Explicitly apply and name an established count-based framework such as DESeq2 (negative binomial Wald test) or edgeR (exact or quasi-likelihood test), reporting the dispersion estimation strategy — Naming the model and software lets readers assess distributional assumptions for count data, judge how overdispersion was handled, and reproduce the analysis; this is standard practice in RNA-seq differential expression reporting
  • TSS classes were defined by a fixed 1-kilobase proximity threshold to ChIP-seq peaks, with overlap counts reported descriptively
    Could also: A permutation or bootstrap test could estimate the expected TSS–peak co-localization under a null model (e.g., randomly shuffled peak positions preserving width and chromosome), yielding an empirical p-value for observed overlap rates — A permutation approach quantifies whether observed co-localization exceeds chance given the genomic distributions of TSSs and peaks, complementing the descriptive percentage-based classification
  • Statistical comparisons of cumulative expression distributions across TSS classes were reported with p-value thresholds via asterisks, without naming the test
    Could also: A two-sample Kolmogorov-Smirnov test or Mann-Whitney U test are standard nonparametric options for comparing full distributions of expression values between groups — Naming the test allows readers to evaluate what aspect of the distributions is being compared (location vs. overall shape), to verify assumptions, and to interpret the statistic in relation to the figures
  • GO enrichment used a threshold-based input (TSSs with FDR < 0.05) without specifying the background gene set or enrichment algorithm
    Could also: Gene Set Enrichment Analysis (GSEA) on a ranked list of all expressed TSSs ordered by differential expression statistic is an alternative that uses the full continuum of evidence rather than a binary cutoff — GSEA avoids dependence on an arbitrary significance threshold for defining the query set and can detect coordinated moderate shifts across many genes; reporting the background set in over-representation analyses similarly helps interpret fold-enrichment estimates
  • Differential expression results were summarized solely by significance category (up/unchanged/down) without effect-size metrics
    Could also: Log2 fold changes and, where replication permits, 95% confidence intervals on those fold changes could be reported alongside FDR-corrected p-values for each differentially expressed TSS — Effect sizes indicate the magnitude of expression change independently of sample size, which is particularly informative for prioritizing candidate TSSs for functional follow-up among the hundreds identified
  • The number of biological replicates for the sciatic nerve Tn5Prime experiment and the primary Schwann cell differentiation assay was not reported in the main text
    Could also: Explicitly stating replicate counts (and ideally a brief power or sample-size justification) per condition is a standard component of sequencing-based differential expression study reporting — Replicate count directly determines the reliability of variance estimates in differential expression models and the generalizability of findings; readers cannot fully evaluate confidence in FDR-corrected results without this information
Software: Not stated

What was reproduced

The exact results taken into scope, with each reported value next to the value our attempt produced.

scope.md — pmid-32770939

Paper: Fogarty EA, Kitzman JO, Antonellis A. "SOX10-regulated promoter use defines isoform-specific gene expression in Schwann cells." BMC Genomics 2020;21:549. PMID 32770939 · PMCID PMC7430845 · DOI 10.1186/s12864-020-06963-7.

Data: GEO GSE139321 — 14 Tn5Prime (5′-anchored, CAGE-like) RNA-seq libraries, rat: 2 sciatic nerve, 3 control + 3 CPT-cAMP primary Schwann cells, 2 parental S16, 4 ΔSOX10 S16. Processed matrix shipped: GSE139321_Schwann_Cell_Tn5Prime_GEO_Processed.txt.gz.

Dependency (previously published) datasets, intersected by the authors:

  • SOX10 ChIP-seq, rat sciatic nerve — GSE64703 (GSE64703_Sox10_peaks_PNS.bed.gz).
  • H3K4me3 ChIP-seq, rat sciatic nerve — GSE84272 (GSE84272_Combined_Sham_H3K4me3_peaks_v2.txt.gz).
  • CAGE mouse tissues — FANTOM5 (used only for Tau scores; lower priority).

Pipeline (from Methods: "Generation and analysis of Tn5Prime…" + "Software…")

FastQC → Cutadapt → STAR (rn5) → SAMtools → Make_CTSS (Takahashi) → Paraclu clustering → featureCounts (per-TSS counts) → edgeR (FDR<0.05). Downstream: BEDTools (TSS→gene within 1 kb same strand; liftOver rn5→mm10 for unannotated; intersect TSS vs SOX10/H3K4me3 peaks within 1 kb). metagene + similaRpeak (AUC permutation), EMBOSS freak/cpgplot, custom perl (motifs), Tau.

IN SCOPE (pipeline-derived, attempted)

ID Result Pipeline Inputs Feasibility
C1 Fig 1a TSS classification: 4993/39706 (12.6%) near H3K4me3∩SOX10 peaks; 2993 loci; 7455 (18.8%) H3K4me3-only; 431 (1.1%) SOX10-only; 26827 (67.6%) neither BEDTools intersect, 1 kb processed TSS file + GSE64703 + GSE84272 peaks HIGH — all 3 files public+processed
C2 Fig 2a: of 4044 candidate TSSs expressed in primary SC, 465 (11.5%) up / 401 (9.9%) down with cAMP edgeR FDR<0.05 processed count matrix (3 ctrl vs 3 cAMP) MED — needs per-sample counts + restrict to 4993 candidate set
C3 Fig 3a: of 4993 candidate TSSs, 265 (5.3%) down / 100 (2.0%) up in ΔSOX10; 3766 (75.4%) unchanged; 862 (17.3%) not expressed edgeR FDR<0.05 processed count matrix (2 parental vs 4 ΔSOX10) MED
C4 Isoform/locus annotation: 739/2993 (25%) ≥2 RefSeq TSSs; 525 (17.5%) multi-TSS distinct CDS (Suppl Table 2) UCSC RefSeq annotation rn5 RefSeq LOW — "manually curated", partly manual

OUT OF SCOPE (wet-lab / manual / not pipeline)

  • Luciferase reporter fold-inductions (ARPC1A 45×, CHN2 10×, DDR1 100×, GAS7 >160×) — wet-lab.
  • Western blot GAS7 ~2.5-fold — wet-lab (ImageJ densitometry).
  • Manual curation of human RefSeq isoform diversity — manual.
  • metagene/similaRpeak AUC permutation p-values — pipeline but secondary; attempt only if core done.
  • Motif/PhastCons/Tau distributions — pipeline but many custom perl steps; lower priority.

Notes on code link

Brief's github.com/adeschen/similaRpeak IS genuinely cited (metagene AUC permutation test) — a third-party tool, valid under P16, but only a secondary result. The authors' own code is github.com/efogarty/Schwann-Cell-SOX10-Promoters (custom perl for motif scans). The CORE reproducible numbers (C1–C3) need neither repo — only BEDTools + edgeR on public processed files.

Figures / tables: Fig 1aFig 2aFig 3a
C1a
Reported
4993 candidate TSSs (Fig 1a)
Reproduced
4993
exact
C1b
Reported
2993 unique loci (Fig 1a)
Reproduced
2993
exact
C1c
Reported
4993/39706 = 12.6% (Fig 1a)
Reproduced
4993 numerator exact; 39706 denominator not deposited
partial
C1d
Reported
all 4993 within 1 kb of H3K4me3 and SOX10 peaks (Methods)
Reproduced
H3K4me3 4993/4993 exact; SOX10 peak file confirmed (99.3% assigned peaks exact in GSE64703 PNS); assignment is locus-level
within tolerance
C2a
Reported
4044 TSSs expressed in primary SC, 81% (Fig 2a)
Reproduced
4044
exact
C2b
Reported
465 up with cAMP, 11.5% (Fig 2a)
Reproduced
465
exact
C2c
Reported
401 down with cAMP, 9.9% (Fig 2a)
Reproduced
401
exact
C3a
Reported
265 down in delSOX10, 5.3% (Fig 3a)
Reproduced
265
exact
C3b
Reported
100 up in delSOX10, 2.0% (Fig 3a)
Reproduced
100
exact
C3c
Reported
3766 unchanged in delSOX10, 75.4% (Fig 3a)
Reproduced
3766
exact
C3d
Reported
862 not expressed in S16, 17.3% (Fig 3a)
Reproduced
862
exact

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

🤖 AI curator · claude (ai-curator room) · v1.0 L1 94/100

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.

🟢1. Data identity
🟢2. Endpoint comparability
🟡3. Location of the main deviation
🟢4. Cause of the deviation
🟢5. Derivability / plausibility
🟢6. Severity of the deviation
🟢7. Core claim
🟡8. Severity of the miss (overall human judgment)
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +1

All nine reported figure counts (Fig 1a/2a/3a) reproduce exactly from the deposited GSE139321 processed table, and the one fully independent computation — a BEDTools 1-kb intersection — recovers H3K4me3 4993/4993 and positively identifies the SOX10 peak file; no fabrication indicators. Two honest caveats sit on the data-availability side, not the authors' integrity: the RPM-only deposit means edgeR cannot be re-run from scratch (so C1a/b, C2, C3 are recounts of the authors' own deposit rather than independent re-execution), and C1c's 39,706 denominator was never deposited. The single non-exact item (C1d) is an explainable locus-level vs per-TSS peak-assignment difference consistent with the paper's alternative-promoter premise. Overall a strong, essentially 1:1 reproduction whose only limitations are deposit completeness and recount-vs-reexecution scope.

🤝
Reproduced automatically — and fairly

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-4-8

Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.

221.2 k
tokens (I/O) · 18.5 M incl. cache
22 min
runtime · 0.01 CPU-h
2.5 GB
peak RAM
1
HPC jobs
hummel
machine