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

Enhancer RNAs stimulate Pol II pause release by harnessing multivalent interactions to NELF.

Nat Commun · 2022
L1 55/100 3/4
Why this verdict

The main results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Nothing in this column.
What did not (or only partly)
  • 🟡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
How its reproducibility compares
55/100
Reproducibility score
1.1 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 15% of all assessed papers rank 986 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

Core in-cell GRO-seq pipeline reproduced END-TO-END on «our HPC» («job»): Trim_galore(default+--polyA) -> STAR mm10/GENCODE-vM25 (8 libs, ~83% unique) -> HOMER de-novo transcript units -> GENCODE overlap; featureCounts -> DESeq2 KCl-vs-CTR; IEG pausing index. RESULTS: C2 activity-induced gene count = 624 vs reported 623 (within-tol, near-exact); C13 mean IEG pausing index falls monotonically CTR 6.32 -> KCl 1.87 = pause release, reproducing Suppl Fig 11d qualitatively; C1 = right pipeline & magnitude but exact 9028 is parameter-sensitive (pooled mrd200=19476, single-lib=9572). The paper's CENTRAL findings are in-vitro biophysics (EMSA/SEC-MALS/XL-MS/anisotropy, Figs 2-4) = OUT OF SCOPE (wet-lab). Code link is the third-party STAR aligner; per brief P16 the documented Trim_galore->STAR->HOMER->DESeq2 pipeline was re-applied to the paper's own data. NOT attempted: C3/C4 (NELFE ChIP), C8 (Exo-seq), C10/C12 (eCLIP) - reproducible in principle (data in deposit) but deferred for compute budget. BLOCKED: eRNA claims C5/C6/C7/C9/C11 depend on an external H3K27ac enhancer annotation not deposited in GSE163113. Dataset GSE163113 profiled: 37/37 runs present (match), grade B (GRO-seq clean & self-contained; eRNA analyses need un-deposited external H3K27ac).

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 50
    assessed: 2026-06-19 ⛓ 89938d0dd3b7
✎ 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
2026-06-26
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19
no human curator yet
Last updated
2026-07-31

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: sonnet
Founding hypothesis

The molecular mechanism by which enhancer RNAs (eRNAs) trigger Pol II pause release was unknown; this study investigates which sequence, structural, and length features of eRNAs enable them to stimulate Pol II pause release by detaching NELF from paused Pol II.

Core claims
  • eRNAs longer than 200 nucleotides that contain unpaired guanosines make multiple, allosteric contacts with NELF subunits -A and -E to trigger efficient NELF release mechanism
  • eRNAs do not exert their function through common structural or sequence motifs and instead populate a wide range of structural spaces finding
  • eRNA-induced NELF dissociation from the paused elongation complex is critically dependent on eRNA length finding
  • Unpaired guanosines, rather than RNA secondary structure per se, are critical for the dissociative effect of eRNAs on the paused elongation complex finding
  • NELF-E-directed eCLIP-seq in mouse primary neurons shows NELF is directly contacted by enhancer-derived eRNAs in vivo finding
  • eRNA-driven NELF release results in transcription activation through more efficient Pol II release from the paused state, as shown in a reconstituted pause release assay finding
  • Exo-seq (5'-end RNA-seq) allows assignment of eRNA transcription start sites with single-nucleotide precision method
  • NELF binding levels correlate with rapid and efficient transcriptional elongation in response to neuronal stimulation finding
Experimental setups
Assay System Perturbation Readout Platform
GRO-seq (global run-on sequencing) mouse cortical primary neurons KCl stimulation (time course) nascent transcription units / eRNA and gene transcription levels
Exo-seq (5'-end RNA-seq) mouse cortical primary neurons KCl stimulation single-nucleotide precision eRNA/mRNA transcription start sites TSScall
SHAPE-MaP (chemical probing + NGS) in vitro transcribed eRNA (1–200 nt) fragments, 39 candidates 1M7 chemical modification SHAPE reactivity / secondary structure ShapeMapper2, RNAstructure
Electrophoretic mobility shift assay (EMSA), radioactive in vitro reconstituted mammalian paused elongation complex (Pol II, DSIF, NELF) on synthetic transcription bubble titration with eRNA fragments, synthetic homopolymer RNAs, and RNA mutants NELF and DSIF dissociation from PEC, apparent Kd
Protein-RNA crosslinking coupled to mass spectrometry in vitro eRNA-bound NELF and paused elongation complexes eRNA binding eRNA contact sites on NELF-A and NELF-E subunits mass spectrometry
eCLIP-seq (enhanced UV crosslinking and immunoprecipitation sequencing) mouse primary neurons, NELF-E-directed neuronal stimulation direct NELF-eRNA contacts genome-wide
ChIP-seq (Pol II 8WG16, CBP, H3K27ac) mouse cortical neurons / publicly available datasets KCl stimulation occupancy at enhancer and gene loci 8WG16 antibody
Reconstituted pause release assay in vitro Pol II transcription system eRNA addition Pol II release from paused state / transcription activation
Key results
  • 1226 intergenic eRNA transcription units identified from GRO-seq, of which 252 were activity-induced >1.5-fold increase (activity-induced definition)
  • 304 eRNAs (281 in replicate 2) showed well-defined 5' ends by Exo-seq; final curated set of 39 high-quality eRNA candidates used for structural/functional analysis n=304; final n=39
  • Median SHAPE reactivities varied broadly across the 39 eRNAs with no shared structural motif, ranging from highly structured (Arc) to highly flexible (Nr4a1-(a)) 0.08 to 0.36
  • eRNA (1–100) fragments showed markedly reduced NELF-dissociation potency compared to (1–200) fragments; (1–50) fragments could not dissociate NELF >10-fold higher apparent Kd for 1–100 vs 1–200
  • Nr4a1-(a) (1–100) fragment dissociated NELF far more effectively than Nr4a1-(b) (1–100) Kd = 2.02 µM vs 0.14 µM
  • Structurally diverse mutants (Arc 96–200, Arc Δstem, Nr4a1-(a) Δloop12) all dissociated NELF equally well, showing structure alone is not determinative
  • Single-stranded low-complexity RNAs lacking guanosines failed to dissociate NELF, while G-containing RNAs (poly(GU)40, poly(GA)48, poly(G2A) variants) efficiently dissociated NELF
  • Guanosines are significantly overrepresented in the 5'-terminal 200 nt of eRNAs p(A/G)=0.018; p(C/G)=0.020; p(U/G)=0.036
Key statistics
  • count 1226 intergenic eRNA transcription units (GRO-seq-defined eRNA transcription units)
  • count 304 eRNAs with well-defined 5' ends (>20 reads per eTSS) (Exo-seq TSS assignment)
  • count 39 eRNA candidates in final test set (SHAPE-MaP structural analysis set)
  • fold_change >1.5-fold increase in eRNA GRO-seq signal (definition of activity-induced eRNAs after KCl stimulation)
  • other median SHAPE reactivity range 0.08–0.36 (structural diversity of eRNAs (Fos e1 lowest, Nr4a1-(a) highest))
  • other apparent Kd = 0.14 µM (Nr4a1-(b) 1–100) vs 2.02 µM (Nr4a1-(a) 1–100) vs 9.70 µM (Nr4a1-(a) 1–50) (NELF dissociation potency of eRNA fragments)
  • fold_change >10x higher apparent Kd for (1–100) vs (1–200) eRNA fragments (length-dependence of NELF dissociation)
  • pvalue p=0.018 (A/G), p=0.020 (C/G), p=0.036 (U/G) (pairwise t-test for guanosine overrepresentation in 5'-terminal 200 nt of eRNAs)

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 paper primarily uses descriptive, qualitative comparisons of gel-based assays (EMSAs) and sequencing-derived metrics (GRO-seq, Exo-seq, SHAPE-MaP, eCLIP-seq) across a curated set of 39 candidate enhancer RNAs, largely reporting fold-differences, apparent binding affinities (Kd) from curve fitting, and one explicit inferential test for nucleotide composition. Group comparisons (e.g., wild-type vs. structural mutants, different eRNA lengths, synthetic homopolymer RNAs) are interpreted mainly from gel patterns and quantified Kd values rather than through a battery of formal significance tests. One pairwise t-test is reported for guanosine overrepresentation in eRNA sequence composition. Data with n and averages are reported for select biochemical assays (e.g., 'mean of two experimental replicates' for EMSA quantification, n = 39 for the SHAPE reactivity boxplot).

Replicationmixed Sample sizen = 39 eRNAs shown as individual data points in the SHAPE reactivity boxplot (Fig. 1d); EMSA Kd curves based on the mean of two experimental replicates GroupseRNA length variants (1-200, 1-100, 1-50), structural mutants (Δstem, Δloop) vs. wild-type, synthetic homopolymer RNAs with varying guanosine content/spacing, and stimulated vs. unstimulated neurons Pairingunclear Randomization/blindingnot stated Dispersionmixed Exact p-valuesyes Confidence intervalsno Multiplicity correctionnone stated
Statistical tests used
Test Applied to n Assumptions
pairwise t-test nucleotide frequency comparison (A vs G, C vs G, U vs G) in the 5′-terminal 200 nt of eRNAs, Fig. 2j not stated
single-site binding model curve fit (apparent Kd determination) quantification of NELF (and Pol II-DSIF complex) dissociation from the PEC across eRNA concentrations, Fig. 2e mean of two experimental replicates not stated
Approaches that could also have been used
  • Guanosine overrepresentation across nucleotide categories (A/G, C/G, U/G) was assessed with three separate pairwise t-tests without a stated multiple-comparisons adjustment.
    Could also: A one-way ANOVA across nucleotide categories followed by a post-hoc test with a correction such as Tukey HSD, Bonferroni, or Benjamini-Hochberg FDR could also be used. — This would jointly test all nucleotide categories at once and control the family-wise error rate that arises from running three related pairwise comparisons on the same dataset.
  • Apparent Kd values for eRNA-driven NELF dissociation were derived from a single-site binding model fitted to the mean of two experimental replicates, without reported variance.
    Could also: Increasing biological/technical replicate number (e.g., n ≥ 3) and reporting the fitted Kd together with a confidence interval or SD/SEM could also be used. — This would let readers gauge the precision and reproducibility of the fitted binding affinity rather than relying on a single point estimate from two replicates.
  • Differences in NELF-dissociation potency between eRNA length variants, structural mutants, and wild-type sequences were primarily interpreted qualitatively from EMSA gel band patterns and fold-differences in Kd.
    Could also: Quantifying replicate band intensities and applying a t-test, or a two-way ANOVA (length × sequence/structure) with appropriate post-hoc correction, could also be used. — This would provide a formal significance assessment (e.g., a p-value) for whether the observed differences between variants exceed what would be expected from assay variability.
  • SHAPE reactivity distributions across 39 eRNAs are summarized with a boxplot of individual data points and reported median values.
    Could also: Reporting the mean ± SD (or 95% CI) alongside the median/IQR, or applying a distribution-based test (e.g., Kruskal-Wallis) to compare reactivity between eRNA subgroups such as IEG vs. non-IEG eRNAs, could also be used. — This would add a complementary quantitative summary and allow a formal comparison between the biologically defined subgroups highlighted in the text (e.g., IEG eRNAs vs. others).
Software: TSScall · ShapeMapper2 · RNAstructure

What was reproduced

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

Scope — pmid-35508485

Paper: Gorbovytska et al. 2022, Nat Commun 13:2425. "Enhancer RNAs stimulate Pol II pause release by harnessing multivalent interactions to NELF." DOI 10.1038/s41467-022-29934-w · PMCID PMC9068813. Data: GEO SuperSeries GSE163113 = subseries GSE163111 (Exo-seq), GSE163112 (SHAPE-MaP), GSE164556 (GRO-seq/eCLIP/ChIP). BioProject PRJNA684920, 37 SRA runs, mouse (mm10). Code link in record: github.com/alexdobin/STAR (third-party aligner, not an authors' analysis repo — methods describe a HOMER/MACS/TSSCall/DESeq2 pipeline applied to this data; per brief P16 running that documented pipeline on the paper's data is an equally valid reproduction).

Nature of the paper

The CENTRAL findings are in-vitro biophysics/biochemistry (multivalent eRNA–NELF interactions): EMSA, SEC-MALS, crosslinking-MS, fluorescence anisotropy, in-vitro pause- release assays — Figs 2,3,4 + Suppl Figs 2–6. OUT OF SCOPE (wet-lab/manual, no pipeline). The in-cell genomics validates the model and IS the reproducible pipeline portion.

IN SCOPE — pipeline-derived results (this study's own data)

# Result (claim) Reported value Fig/loc Data needed Pipeline Tractability
C1 Annotated genes overlapped by GRO-seq de-novo transcript units 9,028 Fig 1 / Methods GRO-seq (8 libs) Trim_galore→STAR mm10→HOMER findPeaks (de-novo TUs)→overlap GENCODE HIGH
C2 Activity-induced genes (KCl-up, FC>1.5) 623 Fig 7b / Methods GRO-seq HOMER analyzeRepeats quant → FC TTX vs KCl HIGH
C3 NELF-bound vs unbound among C2 548 / 75 Fig 7b + NELFE ChIP MACS2 NELFE peaks ∩ TSS MED
C4 NELF-bound split high/mid/low 216 / 133 / 199 Fig 7b + NELFE ChIP signal tertile by ChIP signal MED
C5 Intergenic eRNA transcription units 1,226 Suppl Data 1 GRO-seq + H3K27ac (EXTERNAL) intergenic TUs ∩ H3K27ac peaks PARTIAL*
C6 Activity-induced eRNAs (KCl-up, FC>1.5) 252 Fig 7c / Suppl Data 1 as C5 FC filter on eRNA TUs PARTIAL*
C7 eRNAs NELF-bound / unbound 144 / 108 Fig 7c + NELFE ChIP peak ∩ enhancer PARTIAL*
C8 Exo-seq eRNAs with well-defined 5′-ends (>20 reads/eTSS) 304 (rep1) / 281 (rep2) Methods/Fig1b Exo-seq (3 libs) Cutadapt→SortMeRNA/Bowtie→STAR→TSSCall MED
C9 of which from activity-induced enhancers 86 (rep1) / 79 (rep2) Methods as C8 + enh defn overlap PARTIAL*
C10 eCLIP: pre-mRNAs with ≥1 NELF-E crosslink site 7,242 / 9,028 Fig 5b eCLIP (6) + GRO-seq R2 first-nt crosslink → assign to TUs MED
C11 eCLIP: eRNAs with crosslink sites 240 / 607 Fig 5c as C10 overlap PARTIAL*
C12 eCLIP: fraction of pre-mRNA crosslinks within first 200 nt ~70% Fig 5b text as C10 positional histogram MED
C13 Pausing index pause-release upon KCl (IEGs) metagene/boxplot (Suppl Fig 11) Suppl Fig 11d GRO-seq PI = promoter(−100..+200)/body(+400..+800), HOMER HIGH (qual.)

* PARTIAL = depends on H3K27ac enhancer definition that is NOT in GSE163113 (the enhancer set traces to co-author Kim T-K's neuronal-activity eRNA datasets / external H3K27ac ChIP). Exact counts therefore hinge on an external input whose provenance/version is not pinned in this deposit → reproducible only up to that ambiguity; will be flagged.

Priority (80% floor → keep going)

  1. C1 (9,028 genes) — cleanest integer, exercises core STAR+HOMER path. → MINIMUM.
  2. C2 (623 activity-induced genes) — adds quantification + FC.
  3. C13 pausing-index pause release — the mechanistic in-cell validation.
  4. C3/C4 NELF-bound subdivisions (adds ChIP).
  5. C10/C12 eCLIP crosslink distribution.
  6. C8 Exo-seq TSS counts.
  7. C5–C7,C9,C11 attempted but expected PARTIAL (external H3K27ac dependency).

OUT OF SCOPE (not attempted — wet-lab/manual)

Figs 2,3,4; Suppl Figs 2–6 (EMSA, SEC-MALS

Figures / tables: Fig 1Fig 7bFig 7cFig 1bFig 5bFig 5c
C1
Reported
9028 annotated genes overlapped by GRO-seq de-novo transcript units
Reproduced
pooled 8-lib mrd200=19476 (stranded); single-lib CTR_R4 mrd200=9572
partial
C2
Reported
623 activity-induced genes (KCl-up, RPKM-FC>1.5 & DESeq2 FDR<0.05 both reps)
Reproduced
624 (KCl 30 min timepoint)
within tolerance
C13
Reported
qualitative IEG pause release upon KCl depolarization (Suppl Fig 11d)
Reproduced
mean IEG pausing index CTR 6.32 -> KCl10 3.73 -> KCl30 2.12 -> KCl60 1.87 (monotonic pause release)
within tolerance
C3
Reported
548/75 NELF-bound/unbound
Reproduced
partial
C4
Reported
216/133/199 NELF high/mid/low
Reproduced
partial
C5
Reported
1226 intergenic eRNA TUs
Reproduced
partial
C6
Reported
252 activity-induced eRNAs
Reproduced
partial
C7
Reported
144/108 eRNA NELF-bound/unbound
Reproduced
partial
C8
Reported
304/281 Exo-seq eTSS
Reproduced
partial
C9
Reported
86/79 eTSS from activity enhancers
Reproduced
partial
C10
Reported
7242/9028 eCLIP pre-mRNA crosslinks
Reproduced
partial
C11
Reported
240/607 eCLIP eRNA crosslinks
Reproduced
partial
C12
Reported
~70% pre-mRNA crosslinks in first 200 nt
Reproduced
partial

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 55/100

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)
🤝
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.

121.5 k
tokens (I/O) · 3.7 M incl. cache
16 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.