Estimation of peptide elongation times from ribosome profiling spectra.
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.
- ✓Reported values were directly comparable
- ✓Reported values are derivable from the shared data
- ✓The central claim held under reproduction
- 🟡Could not use the authors’ exact input data
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡The deviation was non-trivial in magnitude
- 🟡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
Described well enough to reproduce 1:1 for the core modelling claim. RiboTimes is the authors' own C++/R (Rcpp) MLE tool and the repo ships the paper's own demo input (data/Demo_input_file.txt, Demo_Data_Set_E_coli, 125 genes, GSE145571). I ran it on «our HPC» and reproduced the Figure 3A/B/C result directly: the two genes the paper names, rpsQ (paper r=0.83 -> repro 0.849) and atpE (paper r=0.81 -> repro 0.820), match within +/-0.02, and the stated 'Pearson r in the 0.7-0.8 range for high-density genes' is reproduced (top-20 density-gene median r=0.761). The Rcpp wrapper segfaulted at dyn.load under conda gcc-15 (SIGSEGV invalid-permissions); I pivoted to the repo's documented standalone C/main.cpp built with gcc-12, which ran cleanly and writes the same quantities to plain text. NOT attempted (hard 20% / out of scope): rebuilding RPF count tables from GSE145571 raw FASTQ (upstream alignment), mupirocin fold-change z-factors (Fig 9A), growth-rate-calibrated absolute elongation times (Fig 6B), and exact full-dataset z-factor/sigma figures (Fig 4A, Supp S5B) -- those use more genes than the 125-gene demo, so on the subset they are only qualitatively comparable (overlapping ranges, sd(log z)=0.40 vs reported 0.61/1.2). No fabrication concern: the reported Fig 3A/B values regenerate directly from the shipped data+code. Verdict: partial, with a clean 1:1 on the headline correlation claim.
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 76assessed: 2026-06-15 ⛓ bd9d00018071
✎ 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
- 2026-06-15
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15no human curator yet
- Last updated
- 2026-09-19
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: sonnetRibosome profiling (RPF) count data at each codon can be decomposed into true peptide elongation times and technical library-generation biases (e.g. nuclease cleavage specificity), and a maximum-likelihood model using local codon context can separate these to yield accurate, bias-corrected elongation time estimates at single-codon resolution.
- ★ A maximum likelihood statistical model using 915 context-defining parameters can neutralize technical biases (e.g. RNase cleavage preferences) in Ribo-Seq data and estimate peptide elongation times at single-codon resolution. method
- ★ RPF counts at an A-site codon follow a Poisson-type distribution with expected value equal to the product of ORF initiation frequency, elongation time, and a context-dependent bias factor. mechanism
- ★ A local sequence context of five codons (spanning A, P, and E ribosomal sites) accounts for the ribosomal dwell time at each A-site codon across the transcriptome. finding
- ★ The method was validated using an independent E. coli dataset treated with the antibiotic mupirocin, which inhibits isoleucyl-tRNA synthetase. finding
- ★ The method was validated using two S. cerevisiae datasets generated with RNases of distinct cleavage specificities (MNase and RNase A), showing consistent bias correction across nucleases. finding
- MNase, commonly used in bacterial Ribo-Seq, cleaves with base-dependent specificity, preferentially before A and U nucleotides, introducing sequence-dependent bias into RPF generation. finding
- Using the 3′-end of MNase-generated RPFs (rather than the 5′-end) to infer the A-site codon improves resolution of bacterial ribosome profiling data due to greater positional precision at the 3′-end. method
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Ribosome profiling (Ribo-Seq) | Escherichia coli B strain AS19, LB medium, OD600 0.5 | none | genome-wide ribosome-protected fragment (RPF) counts per codon | deep sequencing; Bowtie 1.2.2 alignment; fastx-toolkit/cutadapt for read processing |
| Ribosome profiling (Ribo-Seq) | Escherichia coli MG1655, MOPS complete synthetic media | drug (200 μM mupirocin, 10 min, inhibits isoleucyl-tRNA synthetase) vs untreated control | genome-wide RPF counts per codon, comparison of elongation times pre/post drug | — |
| Ribosome profiling (Ribo-Seq) | Saccharomyces cerevisiae | RNase choice: MNase-generated libraries | genome-wide RPF counts per codon | — |
| Ribosome profiling (Ribo-Seq) | Saccharomyces cerevisiae | RNase choice: RNase A-generated libraries | genome-wide RPF counts per codon | — |
- – Model fitting with 915 context parameters via non-linear regression/maximum likelihood successfully fits model-predicted RPF spectra to experimental transcriptome-wide RPF spectra
- – An inner local context of five codons (including A, P, E sites) determines ribosomal dwell time at each A-site codon
- – Method correctly identifies elongation time changes in E. coli upon mupirocin treatment (isoleucyl-tRNA synthetase inhibition)
- – Method neutralizes RNase-specific cleavage biases consistently in S. cerevisiae datasets generated with MNase versus RNase A, which have distinct cleavage signatures
- – MNase-generated RPFs show more variable lengths at 5′-ends than 3′-ends
- count 915 context-defining parameters (number of parameters used in the model to capture local codon context effects on elongation time and bias)
- other OD600 of 0.5 (growth condition at which E. coli AS19 cultures were harvested for ribosome profiling)
- other 200 μM mupirocin for 10 min (antibiotic treatment condition for E. coli MG1655 dataset inhibiting isoleucyl-tRNA synthetase)
- other Bowtie 1.2.2, parameters -l 16 -n 1 -e 50 -m 1 --strata --best (read alignment parameters used for mapping RPFs to the E. coli genome (strain MG1655, U00096.3))
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 paper develops a maximum likelihood (ML) estimation framework to extract peptide elongation times from ribosome profiling (Ribo-Seq) data. RPF counts at each codon position are modeled as Poisson-distributed random variables, and a transcriptome-wide log-likelihood is maximized jointly over 915 context-defining parameters via non-linear regression. The method is applied to original E. coli data and validated against publicly available datasets from E. coli (antibiotic-treated) and S. cerevisiae (two RNases). Results are reported as bias-corrected parameter estimates rather than classical hypothesis-test outcomes.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Maximum likelihood estimation with Poisson likelihood (non-linear regression over transcriptome-wide log-likelihood) | Estimation of 915 codon-context parameters (elongation times and bias factors) from RPF count spectra across all ORFs | 915 context-defining parameters; transcriptome-wide RPF counts from E. coli and S. cerevisiae datasets; exact codon/ORF count not stated in provided text | stated |
-
RPF counts are modeled with a Poisson distribution; the paper itself notes that a Neyman type A distribution (overdispersed relative to Poisson) is the more exact model for sequencing count data after PCR amplification, and simplifies to Poisson only when the product A·q is small↳ Could also: A negative binomial (NB) likelihood — or the full Neyman type A likelihood the authors describe — could also be used to model RPF counts, as these naturally accommodate overdispersion common in sequencing data — When counts are overdispersed (variance > mean), a Poisson assumption underestimates parameter uncertainty; NB adds a dispersion parameter that absorbs extra-Poisson variability, which can yield better-calibrated uncertainty estimates for the elongation-time parameters
-
Parameter uncertainty (standard errors or confidence intervals) for the 915 estimated elongation-time and bias parameters is not reported in the provided text↳ Could also: Profile likelihood confidence intervals, the Fisher information matrix (observed or expected), or parametric bootstrap resampling could also be used to quantify uncertainty around the ML point estimates — Reporting uncertainty bounds alongside point estimates allows readers to distinguish precisely versus imprecisely estimated elongation times and to propagate that uncertainty into downstream biological interpretation
-
Model parameters are estimated by maximizing a frequentist log-likelihood↳ Could also: A Bayesian hierarchical model with priors on elongation times and bias factors (e.g., gamma or log-normal priors) could also be used to jointly estimate parameters and their posterior distributions — A Bayesian approach would naturally provide full posterior uncertainty for each parameter, allow partial pooling across codons or ORFs (shrinkage), and handle low-count positions more gracefully through the prior
-
Biological replicates are described but the provided text does not report how replicate-to-replicate variability is incorporated into the ML fitting or used to assess model reproducibility↳ Could also: Replicate datasets could also be used to compute inter-replicate Pearson or Spearman correlations of estimated elongation times, or to perform a likelihood-ratio or permutation test of model fit across replicates — Formal assessment of replicate concordance for the estimated parameters would provide an empirical measure of reproducibility that complements in-model goodness-of-fit statistics
-
Model selection (e.g., choosing a context window of five codons) is described qualitatively; the statistical criterion used to select context width is not stated in the provided text↳ Could also: Information criteria such as AIC or BIC, or cross-validation on held-out ORFs, could also be used to formally compare models with different context window sizes — A penalized-likelihood or cross-validation criterion would provide a principled, data-driven way to balance model complexity against fit quality, making the chosen context size easier to compare to alternatives
-
RPF counts are normalized per total mapped reads per million (RPM) prior to modeling↳ Could also: Median-of-ratios normalization (as in DESeq2) or trimmed mean of M-values (TMM, as in edgeR) could also be used to normalize read counts across samples before or within the modeling framework — RPM normalization assumes that total mapped read depth is a valid reference across samples; compositional normalization methods are less sensitive to a few highly expressed ORFs dominating the library and are standard in comparative Ribo-Seq analyses
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-33885812 (RiboTimes)
Paper: Estimation of peptide elongation times from ribosome profiling spectra. Tunney? — Dao Duc / Gustafsson et al., NAR 2021. PMID 33885812 · PMCID PMC8136808 · DOI 10.1093/nar/gkab260. Code: https://github.com/gustafGitHub/RiboTimes (master, last push 2021-03-25, C++/R, Rcpp, no license). Data: GEO GSE145571 (E. coli AS19 ribosome profiling) + validation libs (E. coli MG1655, S. cerevisiae).
The tool (what the pipeline does)
RiboTimes is an R wrapper (riboTimesMle.R) around a C++ maximum-likelihood
estimator (C/RiboTimesMLE.cpp, exported runMLE() via Rcpp, cpp11 plugin only —
no Eigen/Armadillo). Input = a text table of ribosome-protected-fragment (RPF)
counts per codon position per gene. Output = a ~40-field R list: experimental vs
modelled selectivity indices (sij_Exper, sij_Model), per-codon z-factors
(zFP, zFP_HAT + their _Sigma), elongation times, and codon statistics.
The repo ships the paper's own demo input data/Demo_input_file.txt (365 KB,
E. coli) plus two yeast 300-gene demo tables — so the codon-count table that feeds
the MLE is provided directly (no upstream read-alignment step needed to reproduce
the modelling result).
IN SCOPE (pipeline-derived, attempted)
The MLE modelling step on the shipped E. coli demo input, and the quantities it emits that the paper reports:
| id | reported claim | paper location | how reproduced |
|---|---|---|---|
| C1 | Pearson correlation between experimental and modelled RPF/selectivity spectra in the 0.7–0.8 range for high-footprint-density genes | Fig 3C / Results | run MleAlgorithm on demo; compute cor(sij_Exper, sij_Model) per gene, look at high-geneRPFdensity genes (rpsQ, atpE named in Fig 3A/B) |
| C2 | σ ≈ 0.61 for bias-free elongation parameters; σ ≈ 1.2 for total context factors | Supp Fig S5B | from the z-factor outputs (zFP_HAT bias-free vs zFP context) spreads / _Sigma fields |
| C0 | the pipeline itself runs on the paper's data and regenerates the described output structure + Fig 2/3 (plot_figures.R) |
repo demo | structural reproduction (compiles, runs, produces 40-field list + figure PDFs) |
Primary, cleanest claim = C1 (unambiguous computation from two emitted vectors). C2 is secondary/exploratory (figure-only value, field mapping less certain).
OUT OF SCOPE (not attempted, why)
- Upstream read processing (FASTQ → RPF count tables from GSE145571 raw reads): not needed — the count tables are shipped; reproducing the aligner step is a separate, much larger pipeline and is the "hard 20%".
- Mupirocin fold-change z-factors (Fig 9A: 13×/16×/8.5×): requires the MG1655 treated/untreated libraries processed into the tool's input format (not in the shipped demo) — out of scope for the 80/20.
- Absolute elongation times calibrated by growth rate (Fig 6B): depends on doublingTime calibration + full dataset; demo-subset values not 1:1 comparable.
- Wet-lab / manual content: none relevant.
Reproduction model
Third-party-vs-own: this is the authors' own repo, applied to the authors' own shipped demo data — a direct 1:1 reproduction of the modelling step.
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.
Running the authors' own RiboTimes MLE on their shipped E. coli demo reproduces the headline Fig 3A/B correlations 1:1 (rpsQ 0.83->0.849, atpE 0.81->0.820, both within +/-0.02) and the stated 0.7-0.8 band for high-density genes (median r=0.761). The only deviations are on secondary full-dataset figures (S5B sigma 0.61/1.2 -> sd(log z)=0.40; Fig 4A edge z-factor ranges), and these sit on our side — a consequence of using the 125-gene demo subset rather than rebuilding count tables from raw FASTQ — not an authors' defect. No fabrication concern; values are derivable from the shared data. Overall a solid reproduction with explainable, scope-driven secondary deviations.
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 [email protected].
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.