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

Translation affects mRNA stability in a codon-dependent manner in human cells.

Elife · 2019
L1 87/100 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 🟡
Minor / cosmetic deviation
+1 pts
From: Q4 · Cause of the deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +2
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • 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
  • 🟡A deviation was attributed to the published material
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
87/100
Reproducibility score
0.7 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 72% of all assessed papers rank 301 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 via a deterministic route. The paper's headline pipeline result is the Codon Stability Coefficient (CSC, Fig 1B) = Pearson correlation, across genes, between each codon's CDS frequency and mRNA stability. I recomputed CSC for all 61 sense codons from the paper's shipped per-gene decay rates (Fig1 source data 1, xlsx) + Ensembl GRCh38 r110 human CDS (longest CDS per gene), and compared to the reported CSC (Fig1 source data 2, csv). Result: HeLa-endo R=0.992, RPE-endo R=0.997, K562-SLAM R=0.999, 293T-endo R=0.919 (244 codon data points total); near-exact except 293T which is noisier. The paper's 'CSC conserved across cell lines' claim (Fig1C) also reproduces (cross-approach R 0.842-0.957 vs reported 0.840-0.960). The small residual (per-codon MAD ~0.005) is explained by using one representative transcript per gene rather than the paper's exact annotation. FABRICATION CHECK: none -- every reported CSC is faithfully derivable from the shipped decay rates; the only quirk is an inconsistent decay_rate sign convention across the 6 sheets (293T sign-inverted vs HeLa/RPE; SLAM stores a raw negative slope), a documentation issue not a data problem. NOT ATTEMPTED (the hard ~20%, per brief): (1) re-running slamdunk v0.3.3 on the raw K562 SLAM-seq fastqs (hg38 + 3'UTR BED + alignment) to regenerate the SLAM half-lives from reads -- heavy plumbing for incremental value once CSC is reproduced; the K562-SLAM CSC was reproduced from the shipped SLAM half-lives instead. (2) ORFome CSC -- clone/ORF ID space not mappable to Ensembl CDS without the ORFome library sequences. Verdict: PARTIAL but strong -- the central codon-dependent-stability result reproduces near-exactly across 4 datasets and 244 data points.

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 87
    assessed: 2026-06-14 ⛓ b0bca0958fe3
✎ 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
2026-06-14
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15
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

Does translation affect mRNA stability in a codon-dependent manner (codon optimality) in human cells, and is the regulatory information encoded in codons rather than nucleotides?

Core claims
  • Translation strongly affects mRNA stability in a codon-dependent manner in human cells, with specific codons stabilizing or destabilizing mRNAs. finding
  • The regulatory information affecting mRNA stability is encoded in codon identity, not in nucleotide composition. finding
  • Stabilizing (optimal) codons tend to be associated with higher tRNA levels and higher charged/total tRNA ratios. mechanism
  • mRNAs enriched in destabilizing codons tend to have shorter poly(A)-tails, but the poly(A)-tail is not required for codon-mediated mRNA stability. finding
  • The codon-mediated effect depends on translation and is modulated by the number of ribosomes loaded onto an mRNA (e.g. via viral infection or specific UTRs). mechanism
  • The codon stability coefficient (CSC) approach quantifies each codon's effect on mRNA stability as the Pearson correlation between codon occurrence and mRNA stability. method
  • An ORFome vector-based library (~17,000 coding sequences with shared UTRs/promoter and 3'UTR barcodes) measures coding-sequence regulatory activity independent of UTRs. resource
  • Codon optimality codes are conserved across human cell lines, zebrafish, and mouse embryonic stem cells. finding
Experimental setups
Assay System Perturbation Readout Platform
Time-course mRNA-seq (endogenous decay profiles) 293T, HeLa, RPE human cell lines Actinomycin D (transcription block) mRNA decay/stability over time, CSC per codon Actinomycin D 5 μg/ml
ORFome decay profiling (barcoded exogenous mRNA-seq) 293T and K562 cells Lentiviral ORFome library + Actinomycin D, puromycin selection Exogenous mRNA decay via 3'UTR barcodes, CSC ~17,000-vector ORFome library
SLAM-seq (metabolic labeling, orthogonal-chemistry RNA-seq) Human cells incl. K562 4-thiouridine (s4U) feed/chase, no transcription block mRNA half-life via T-to-C conversion, CSC
1nt-frameshift fluorescent reporter (qPCR + flow cytometry) 293T (also HeLa, RPE, K562) Single-nucleotide frameshift converting optimal to non-optimal CDS; Actinomycin D Reporter mRNA stability and mCherry/GFP fluorescence intensity
'Extreme' premature-stop-codon reporter (qRT-PCR) Human cells (293T) Single mutation creating premature stop codon upstream of optimal/non-optimal region; Actinomycin D Relative reporter mRNA stability
tRNA level and charged/total tRNA ratio measurement Human cells none tRNA abundance and aminoacylation (charging) ratios vs codon optimality
Poly(A)-tail length measurement Human cells none Poly(A)-tail length vs codon composition
m6A target mRNA decay comparison Human cells (endogenous vs ORFome) none (predicted m6A targets vs control) Scaled mRNA decay of m6A targets vs controls
Key results
  • Optimal codons show positive correlation and non-optimal codons negative correlation with mRNA stability (CSC) across human cell lines.
  • CSC scores correlate well across 293T, HeLa, RPE human cell types and zebrafish.
  • ORFome-derived CSC scores correlate with endogenous mRNA CSC scores.
  • Endogenous m6A target mRNAs are more unstable than controls, but this difference is largely lost in ORFome-derived mRNAs (lacking 3'UTR elements).
  • SLAM-seq half-lives strongly correlate with TimeLapse-seq in K562 and SLAM-seq CSC correlates with mouse ESC data.
  • Optimal-codon frameshift reporter mRNA is more stable and shows higher mCherry intensity than its non-optimal counterpart, with no change in GFP control.
  • Optimal-codon reporter shows higher RNA level than non-optimal counterpart 24 hr post-transfection (homeostatic effect).
  • Codon-composition effects are independent of frameshift nucleotide identity, fluorescent protein, and coding sequence used.
Key statistics
  • count ~17,000 expression vectors/coding sequences (ORFome library size)
  • other 20-nucleotide unique barcode (barcode length in 3'UTR of ORFome vectors)
  • pvalue ***p<1e-16 (Spearman correlation significance for CSC across cell lines (Fig 1C))
  • pvalue *p<1e-50 (Kolmogorov-Smirnov test on mRNA level distribution (Fig 1E))
  • pvalue *p<0.05, **p<0.01, ***p<0.005 (Unpaired t-test for reporter stability panels (Fig 2B))
  • other s4U fed every 3 hr for 24 hr (SLAM-seq labeling protocol)
  • other Cycloheximide 2 μg/ml; Actinomycin D 5 μg/ml (drug concentrations for translation/transcription block)
  • count 61 codons (codons scored for stability properties (CSC))

Statistical methods review

Model: opus

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 measures codon-dependent mRNA stability in human cells using three independent approaches (Actinomycin D time-course mRNA-seq, an exogenous ORFome reporter library, and SLAM-seq metabolic labeling) across four cell lines. The central metric is the codon stability coefficient (CSC), defined as the Pearson correlation between each codon's occurrence and mRNA stability, and agreement across methods/cell types is assessed with Pearson and Spearman rank correlations. Distributional and group comparisons are reported with Kolmogorov-Smirnov tests, Wilcoxon rank-sum tests, linear models, and unpaired t-tests for reporter experiments, with significance shown as p-value thresholds.

Replicationmixed Sample sizeNumber of genes (n) reported for some genomic comparisons (e.g. Figure 1F); reporter experiments mention triplicates (Figure 1—figure supplement 1E) and duplicates (supplement 1F); formal power analysis not described Groupsoptimal vs non-optimal codon-enriched mRNAs/reporters; m6A targets vs control mRNAs; methods/cell lines vs each other Pairingunpaired Randomization/blindingnot stated Dispersionunclear Exact p-valuesno Effect sizesno Confidence intervalsno Multiplicity correctionnone stated
Statistical tests used
Test Applied to n Assumptions
Pearson correlation coefficient (used to compute CSC) Figure 1A/B — correlation between codon occurrence and mRNA stability; also Figure 1—figure supplement 1F,G,H,I correlations between methods/datasets not stated
Spearman rank correlation Figure 1C and Figure 1—figure supplement 1B,C — correlations of CSC scores across cell lines, methods, and vs codon usage not stated
Two-sample Kolmogorov-Smirnov test Figure 1E (mRNA level distributions, 293T/K562) and Figure 1F (m6A-target vs control mRNA decay) number of genes (n) indicated in Figure 1F not stated
Linear model Figure 1F — estimating the difference in scaled mRNA decay between m6A targets and controls not stated
Wilcoxon rank-sum test Figure 1—figure supplement 1D — nucleotide transition ratio with/without s4U treatment not stated
Unpaired t-test Figure 2B reporter decay panels (stated 'Unpaired t.test for all the panels'); related reporter comparisons (2C, 2D) not stated
Approaches that could also have been used
  • Codon influence on stability was quantified as the CSC using the Pearson correlation between codon occurrence and mRNA stability.
    Could also: A Spearman rank correlation, or a multiple regression / linear model with all 61 codons entered jointly, could also be used to derive per-codon coefficients. — A rank-based measure is robust to outliers and non-linearity, and a joint regression can account for correlated codon usage across transcripts, which would complement the single-codon Pearson approach.
  • Reporter group comparisons were assessed with unpaired t-tests across several panels.
    Could also: A single ANOVA (or mixed-effects model) with a post-hoc correction, or a non-parametric Mann-Whitney U test, could also be applied. — An ANOVA with post-hoc adjustment controls the family-wise error rate across the multiple related comparisons, while a non-parametric test relaxes the normality assumption that is helpful with small sample sizes.
  • Significance was reported using threshold p-values (e.g. p<0.05, p<1e-16).
    Could also: Reporting exact p-values together with effect sizes and 95% confidence intervals could also be done. — Exact p-values and effect sizes with intervals convey both the magnitude and precision of an effect, which adds interpretive information beyond a significance threshold.
  • Distributional differences (e.g. m6A targets vs controls) were evaluated with the Kolmogorov-Smirnov test alongside a linear model.
    Could also: A Wilcoxon rank-sum test or an explicit effect-size estimate (e.g. difference in medians with a CI) could also summarize the shift. — These approaches directly quantify the location shift and its uncertainty, complementing the KS test which is sensitive to any distributional difference.
  • Replicate structure is described in places as triplicates or duplicates without a formal sample-size/power justification.
    Could also: An a priori power calculation or explicit statement of biological versus technical replication could also be reported. — Documenting the replication basis and expected effect size helps readers interpret the precision of the estimates and the n underlying each test.

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.

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.

Citations
314
Impact: very high
Foundation confidence
None of its references are in our reproducibility record yet — its foundation cannot be assessed.
Topics

Assessed papers, coloured by verdict. Click a node to open it.

Built on (assessed references) (0)
  • No assessed neighbours yet — the network grows as more papers are assessed.
Cited by (assessed papers) (1)

What was reproduced

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

Scope — pmid-31012849

Paper: Wu Q, Medina SG, Kushawah G, et al. Translation affects mRNA stability in a codon-dependent manner in human cells. eLife 2019;8:e45396. PMID 31012849.

Named code: https://github.com/t-neumann/slamdunk (slamdunk, the SLAM-seq aligner/quantifier; v0.3.3 used by the paper). This is a third-party tool applied to the paper's data — per the brief, equally valid as own-code.

Data: GEO GSE126523 (SuperSeries; SubSeries GSE126520 endogenous decay 293T/HeLa/RPE, GSE126521 ORFome decay, GSE126522 K562 SLAM-seq). SRA PRJNA522258.

Pipeline-derived results (what the paper computes)

  1. mRNA decay rates / half-lives per gene, by exponential fit of expression (or labeled-fraction for SLAM-seq) over a 0–6 h time course, per cell line / approach. → Figure 1—source data 1 (xlsx, 6 sheets: 293T-endo, HeLa-endo, RPE-endo, 293T-ORFome, k562-ORFome, k562-SLAM-seq).
  2. Codon Stability Coefficient (CSC) per codon = Pearson correlation, across genes, between a codon's occurrence frequency in the CDS and the mRNA's stability. This is the paper's headline quantitative result ("codon- dependent"). → Figure 1—source data 2 (csv): CSC per codon × approach.
  3. CSC conservation across cell lines / methods and vs prior data (Fig 1B/1C): CSC vectors correlate strongly across approaches.

IN SCOPE (attempted)

  • Recompute CSC (result 2) from the shipped decay rates (result-1 xlsx) + human CDS (Ensembl GRCh38 r110), and compare all 61 sense codons to the reported CSC (Fig 1—source data 2), for the 4 approaches keyed by Ensembl gene IDs: 293T-endo, HeLa-endo, RPE-endo, K562-SLAM. (244 codon data points.)
  • Recompute CSC cross-approach conservation (result 3) and compare to the reported-column cross-correlations.

OUT OF SCOPE / not attempted (the hard ~20%, by design)

  • Re-running slamdunk on the K562 SLAM-seq fastqs (full map→filter→snp→count on hg38 with a 3′UTR BED) to regenerate the SLAM half-lives from raw reads. Heavy (multi-GB fastq + genome index + annotation plumbing) for incremental value once the headline CSC is reproduced via a deterministic route. The K562-SLAM CSC is still reproduced here (from the shipped SLAM half-lives).
  • ORFome CSC (293T-ORFome, k562-ORFome): the ORFome sheets key on a clone/ORF ID space (not Ensembl gene IDs), so the quick CDS-based recompute cannot map them without the ORFome library sequences. Skipped.
  • Wet-lab / reporter results (Figs 2–5 source data are oligo/reporter sequences, not pipeline outputs) — out of scope.

Caveat affecting exactness

The paper's exact transcript set per gene is not pinned in the source data; I use one representative CDS per Ensembl gene (longest CDS, Ensembl GRCh38 release-110). This explains the small residual (per-codon MAD ~0.005) vs an exact transcript match; it does not affect the across-codon agreement (R up to 0.999).

Figures / tables: Fig 1BFig 1Fig 1C
C1_293T_endo_CSC
Reported
Fig1 source-data2 CSC col 293T_endo (61 codons)
Reproduced
Pearson R=0.919, MAD=0.042 vs reported (293T decay_rate sign-inverted vs HeLa/RPE)
within tolerance
C2_HeLa_endo_CSC
Reported
Fig1 source-data2 CSC col HeLa_endo (61 codons)
Reproduced
Pearson R=0.992, MAD=0.0061 vs reported
exact
C3_RPE_endo_CSC
Reported
Fig1 source-data2 CSC col RPE_endo (61 codons)
Reproduced
Pearson R=0.997, MAD=0.0046 vs reported
exact
C4_K562_SLAM_CSC
Reported
Fig1 source-data2 CSC col K562_SLAM (61 codons)
Reproduced
Pearson R=0.999, MAD=0.0035 vs reported (from shipped SLAM half-lives)
exact
C5_CSC_conservation
Reported
Fig1C: CSC conserved across cell lines, cross-R 0.840-0.960
Reproduced
cross-R 0.842-0.957 (matches within ~0.01-0.04)
within tolerance
C6_ORFome_CSC
Reported
Fig1 source-data2 CSC cols 293T_ORFome, K562_ORFome
Reproduced
not attempted (ORFome sheets key on clone/ORF IDs, not Ensembl gene IDs)
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 87/100

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 🟡
Minor / cosmetic deviation
+1 pts
From: Q4 · Cause of the deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +2

The paper's headline output (the Codon Stability Coefficient, Fig1B) reproduces near-exactly from the authors' shipped decay rates + Ensembl GRCh38 CDS: R=0.992/0.997/0.999 (HeLa/RPE/K562) and 0.919 (293T, noisier), and the CSC-conservation claim (Fig1C) reproduces within ~0.01–0.04. The only deviations sit on the input/preprocessing side and are minor: our choice of one representative transcript per gene (MAD ~0.005, larger 0.042 for 293T) and an inconsistent cross-sheet decay_rate sign convention (a documentation issue, not fabrication). Two items were not attempted (ORFome CSC, re-running slamdunk on raw fastqs), so coverage is partial but the core conclusion holds 1:1.

🤝
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.

161.1 k
tokens (I/O) · 6.8 M incl. cache
15 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.