Frequency and patterns of ribonucleotide incorporation around autonomously replicating sequences in yeast reveal the division of labor of replicative DNA polyme
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.
- 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
Re-analysis paper applying the third-party Ribose-Map tool (P16-valid, agombolay/ribose-map @ fff581bf) to public rNMP-seq data. IN SCOPE = per-library genome-wide rNMP composition (rA/rC/rG/rU %). Ran the full Ribose-Map pipeline (umi_tools extract NNNNNNXXXNN + TCA demux -> bowtie2 to sacCer2 -> umi_tools dedup -> coordinate -> composition) on SRR11364933 (rnh201-null ribose-seq, authors' own data, repo's bundled lib/SRR11364933.config) on «our HPC» compute node n093. Alignment 74.72%, 692,596 nuclear rNMP coordinates. RESULT: composition rC 54.67% >> rG 28.26% > rU 9.51% > rA 7.55%. C2 (rC most abundant) REPRODUCED cleanly (exact, categorical). C1 (rU least abundant) only PARTIAL: the literal claim is strand/ARS-resolved (authors' non-public code, out of scope); the reproducible genome-wide composition puts rA, not rU, as the strict minimum on nuclear chromosomes (rU is least only on mtDNA) -- flagged as a claim-scope difference for the human reviewer, NOT a fabrication assertion. OUT OF SCOPE (not attempted): ARS leading/lagging ratios, regression vs firing time/efficiency, Monte-Carlo simulations, dinucleotide preference p-values -- all produced by the authors' non-public custom analysis code, not reproducible from any shipped artifact. Note: env build initially failed twice (deprecated r-channel/meme hang; then umi_tools=1.0.0 pin pulled a broken pysam under free-threaded Python 3.14) -- fixed by pinning python=3.10 + umi_tools>=1.1.
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 50assessed: 2026-06-18 ⛓ 4152748cfb28
✎ 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
- 2026-06-25
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19no 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: sonnetThe paper tests whether ribonucleotide (rNTP) incorporation into yeast genomic DNA follows specific sequence/positional patterns around replication origins (ARSs), and seeks to confirm—using wild-type as well as mutant polymerase cells—the model that DNA Pol δ initiates leading-strand synthesis before handing off to Pol ϵ.
- ★ rNTP incorporation is preferentially found on the leading strand in yeast cells expressing wild-type replicative DNA polymerases finding
- ★ The leading/lagging-strand ratio of rNTP incorporation changes dramatically within the first 1,000 nucleotides from ARSs, reflecting the Pol δ-to-Pol ϵ handoff during early leading-strand synthesis finding
- ★ The pattern of rNTP incorporation (base and dinucleotide preferences) is markedly distinct between leading and lagging strands in both wild-type and mutant polymerase cells finding
- ★ Distinct rNTP incorporation signatures of Pol δ and Pol ϵ provide a new computational approach to track the division of labor of replicative polymerases at the replication fork mechanism
- ★ A simulation model incorporating polymerase-specific rNTP incorporation rates and tract lengths reproduces the observed rate changes on leading and lagging strands around ARSs method
- Log-leading/lagging ratio of rNTP incorporation decreases with ARS firing time across multiple independent library types finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| ribose-seq (rNMP mapping/sequencing) | Saccharomyces cerevisiae (BY4741/BY4742 strains), wild-type and mutant Pol/RNase H2 genotypes | rnh201-null and/or polymerase steric-gate mutants (pol1, pol2, pol3) vs wild-type | genomic location and count of embedded rNMPs relative to ARSs (leading vs lagging strand) | Ribose-Map bioinformatics toolkit; sacCer2 reference genome |
| emRiboSeq (rNMP mapping/sequencing) | Saccharomyces cerevisiae, rnh201-null with mutant polymerase alleles (e.g., pol2-M644G) | rnh201-null; catalytic polymerase mutants | rNMP incorporation sites/counts around ARSs, leading/lagging strand bias | Ribose-Map bioinformatics toolkit; sacCer2 reference genome |
| RHII-HydEn-Seq (rNMP mapping/sequencing) | Saccharomyces cerevisiae (E134, YFP17, Δl(-2)l-7BYUNI300 strain backgrounds) | rnh201-null with wild-type or mutant (pol1, pol2, pol3) alleles | rNTP incorporation sites (BigWig/BED), leading/lagging ratios and dinucleotide patterns around ARSs | bigWigToBedGraph plus custom script; L03 reference genome |
| Computational simulation of rNTP incorporation | In silico model of single and combined (n=400) yeast ARSs | varied polymerase tract lengths and incorporation rates (wild-type vs mutant Pol δ/ϵ) | simulated rNTP incorporation rate profile across leading/lagging strands | custom script (Data Availability) |
- ▲ Wild-type polymerase yeast cells show a preference for rNTP incorporation on the leading strand around ARSs
- – Leading/lagging ratio changes sharply within the first 1,000 nt downstream of ARSs, marking the Pol δ-Pol ϵ handoff
- ▼ Log-leading/lagging ratio of rNTP incorporation decreases with increasing ARS firing time across wild-type and rnh201-null ribose-seq, emRiboSeq, and RHII-HydEn-seq libraries coefficients -0.0061 to -0.0321
- – Simulated and observed PPB curves show increasing rNTP incorporation rate on leading strand and decreasing/steady rate on lagging strand at replication start, matching wild-type polymerase model
- – rNTP incorporation base/dinucleotide preference patterns differ distinctly between leading and lagging strands in both wild-type and mutant genotypes
- correlation coefficient = -0.0061 (log-leading/lagging ratio vs ARS firing time, wild-type RNase H2 ribose-seq libraries)
- correlation coefficient = -0.0321 (log-leading/lagging ratio vs ARS firing time, rnh201-null ribose-seq libraries)
- correlation coefficient = -0.0082 (log-leading/lagging ratio vs ARS firing time, rnh201-null emRiboSeq libraries)
- correlation coefficient = -0.0306 (log-leading/lagging ratio vs ARS firing time, rnh201-null RHII-HydEn-seq libraries (L03 genome))
- fold_change mutant Pol ϵ ~5-fold, mutant Pol δ ~10-fold of wild-type incorporation rate (estimated from rNMP counts around ARSs in RHII-HydEn-seq libraries)
- other Pol α 1 rNTP/625 nt; Pol δ 1 rNTP/5,000 nt; Pol ϵ 1 rNTP/1,250 nt (in vitro average rNTP incorporation rates of replicative polymerases)
- count 410 confirmed ARSs (sacCer2); 276 with known firing time (139 early, 137 late) (ARS annotation set used for ribose-seq/emRiboSeq analysis)
- count 465 ARSs with firing time (233 early, 232 late) (ARS annotation set used for RHII-HydEn-seq analysis)
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.
This computational genomics study analyzed published rNMP-sequencing datasets (ribose-seq, emRiboSeq, RHII-HydEn-seq) to characterize ribonucleotide incorporation patterns around autonomously replicating sequences (ARSs) in yeast. The core analytical framework modeled per-library rNMP counts on each strand as Poisson-distributed and applied maximum likelihood estimation (MLE) to derive leading/lagging strand incorporation rate ratios; linear regression coefficients were reported to describe the relationship between ARS firing time and log-transformed strand ratios. A simulation based on known polymerase tract lengths and incorporation rates was used to model expected patterns. Results were reported primarily descriptively — using mean ratios, IQR-based error bars, standard deviations, and regression coefficients — without formal null-hypothesis significance tests or p-values.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Maximum likelihood estimation (MLE) under a Poisson count model | Estimation of per-strand rNTP incorporation rates (a, b) and their leading/lagging ratio (θ) across ARS flanks, pooled within genotype × technique groups | N=6 wt-RNase H2 ribose-seq libraries; N=8 rnh201-null ribose-seq; N=5 emRiboSeq; N=4 RHII-HydEn-seq; n=276 or 465 ARSs with known firing time depending on reference genome | stated |
| Linear regression (regression coefficient only) | Scatter plots of log-leading/lagging ratio vs ARS firing time, per library group (Figures 1D, 1E) | n=276 ARSs (sacCer2 datasets); n=465 ARSs (RHII-HydEn-seq / L03); library N as above | not stated |
| Descriptive comparison of mean leading/lagging ratios | Bar graphs comparing strand bias across genotypes and early vs late-firing ARSs (Figure 1C) | N=4–8 libraries per group; n=139/137 early/late ARSs (sacCer2) or n=233/232 (L03) | not stated |
| Stochastic simulation (custom script, Poisson/Normal deviation model) | Modeling expected rNTP incorporation rate profiles on leading and lagging strands for single and combined ARSs (Figures 2A–D) | n=400 ARSs simulated; ARS annotation deviation modeled as Normal(SD=1000 nt); Pol δ tract length uniform over 500–4000 nt | stated |
-
The leading/lagging strand ratio was estimated via MLE and reported as a point estimate, without a confidence interval or formal test of whether the ratio differs from 1↳ Could also: A Poisson rate-ratio test or exact Poisson confidence interval for θ = a/b could be derived from the same MLE framework and reported alongside the point estimate — Confidence intervals on the strand-bias ratio would allow readers to assess uncertainty around each estimate and judge whether values confidently exceed or fall below 1, which is the biological null of equal incorporation on both strands
-
Inter-library variability within a genotype group is summarized with 1.5 IQR error bars, based on N=4–8 libraries↳ Could also: Standard deviation (SD) or 95% confidence interval of the mean could also be used to convey variability across libraries — For small library counts (N=4–8), SD and CI directly characterize the precision of the estimated group mean; IQR is most informative for summarizing the spread of larger or skewed distributions
-
Comparisons between genotype groups and ARS subgroups were made descriptively by visual inspection of mean ratios in bar graphs, without formal statistical tests↳ Could also: A non-parametric Wilcoxon rank-sum test (or permutation test) across libraries within each group could also be used to formally test whether strand ratios differ between conditions — Formal tests provide a quantified probability that observed differences arise by chance, complementing descriptive comparisons — especially when library counts are small and patterns are compared across many genotypes simultaneously
-
The association between ARS firing time and log-leading/lagging ratio was described by a single regression coefficient, with no p-value, confidence interval on the slope, or goodness-of-fit statistic reported↳ Could also: Reporting the 95% CI and p-value for the regression slope, or a Spearman rank correlation with CI, could also characterize this association — A slope CI and significance test allow readers to evaluate whether the observed negative trend is distinguishable from zero; Spearman correlation is additionally robust to non-linear monotonic relationships between firing time and strand bias
-
Multiple comparisons across genotypes, ARS subgroups, library types, and genomic bins were examined without any multiple-testing framework↳ Could also: If formal tests were added, a Benjamini–Hochberg FDR procedure could also be applied across the family of comparisons to control the expected proportion of false discoveries — Transparently acknowledging the number of comparisons helps readers calibrate confidence in individual observed patterns, even in a primarily descriptive study
-
rNMP incorporation preference (heatmaps) was quantified as normalized frequency relative to genomic dNMP background, without reporting a measure of the magnitude of enrichment or its uncertainty↳ Could also: A log2 odds ratio or fold-enrichment relative to background, with a bootstrap confidence interval, could also quantify the degree of rNMP-type preference at each position — An effect-size measure such as a log odds ratio provides an interpretable, symmetric scale for enrichment/depletion and facilitates direct comparison of preference magnitudes across datasets and conditions
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-34551434
Paper: Xu P, Storici F. Frequency and patterns of ribonucleotide incorporation around autonomously replicating sequences in yeast reveal the division of labor of replicative DNA polymerases. Nucleic Acids Res 2021;49(18):10542–10557. PMID 34551434 · PMCID PMC8501979 · DOI 10.1093/nar/gkab801.
Code: https://github.com/agombolay/ribose-map (third-party tool: Ribose-Map,
Gombolay et al.) — pinned commit fff581bf28ff3e6c3970da99ccfb29df8c8de439
(last commit 2022-04-03). This is P16-valid: the paper applies an existing
third-party GitHub tool (Ribose-Map) to public rNMP-sequencing data. The paper's
own custom downstream analysis code (ARS-flank binning, leading/lagging ratio
maximum-likelihood estimation, linear regression vs firing time, Monte-Carlo
simulations) is NOT in a public repo — only the Ribose-Map preprocessing tool
is.
Nature of paper: a re-analysis / meta-analysis. The authors do NOT generate all data; they reuse published rNMP-seq libraries from 4 BioProjects (only the ribose-seq set PRJNA613920 is the authors' own) and run Ribose-Map on them, then apply custom code for the ARS-centric statistics.
Datasets the paper relies on
| accession | technique | role | n libraries (paper) |
|---|---|---|---|
| PRJNA613920 | ribose-seq | authors' own (brief's named accession) | 15 |
| PRJNA271170 | emRiboSeq | reused (Koh/Clausen) | 15 |
| PRJNA517710 | RHII-HydEn-seq | reused | 17 |
| PRJNA261234 | pol2-mutant ribose-seq | reused | — |
| Total analysed: 47 libraries. Reference: sacCer2 (SGD, 2008) for | |||
| ribose-seq/emRiboSeq. |
IN SCOPE (Ribose-Map pipeline-derived, reproducible)
Ribose-Map modules produce, per library, the genomic coordinates and counts of incorporated rNMPs and the rNMP composition (normalized % of rA/rC/rG/rU). These feed every figure. The directly-reproducible, low-ambiguity pipeline outputs:
- S1 — rNMP composition of an rnh201-null ribose-seq library. Run
alignment→coordinate→composition on SRR11364933 (= sample
rnh201-E134-RE2-FS141, PRJNA613920; the repo's own bundled example config) with sacCer2.- Paper claim C1: "rU is always the least abundant rNTP incorporated on both the leading and lagging strands." → check rU is the minimum of {rA,rC,rG,rU}.
- Paper claim C2: "rC is the most abundant rNMP in rnh201-null ribose-seq and RHII-HydEn-seq libraries." → check rC is the maximum.
- Pipeline: Ribose-Map alignment (umi_tools extract NNNNNNXXXNN, TCA barcode demux, bowtie2 to sacCer2, umi_tools dedup) → coordinate (ribose-seq rNMP position transform) → composition (background-normalized %).
OUT OF SCOPE (not attempted)
- Leading/lagging strand ratios around ARS, the 500-nt binning, regression coefficients vs firing time/efficiency (Fig 1–3), Monte-Carlo simulations (Fig 2): produced by the authors' non-public custom analysis code — not reproducible from any shipped artifact (possible-fabrication-unverifiable note in AUDIT, not a fabrication claim).
- Dinucleotide preference p-values (Fig 4–6): same — custom code.
- Wet-lab library construction, strain genotyping: experimental, out of scope.
Approach
Reproduce the Ribose-Map composition for the authors' own rnh201-null ribose-seq library (most direct: same tool, authors' own data, repo's own example config) and profile all 4 datasets. The ~80% floor = the categorical composition claims (C1/C2) on at least one library; then extend to more libraries / dinucleotide ordering if feasible.
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.
This is a preliminary, incomplete reproduction of a re-analysis paper: only the peripheral categorical composition claims (C1 rU least abundant, C2 rC most abundant in rnh201-null) were in scope, and even those are PENDING («job» still running, claims.tsv empty, agreement.json not-run-yet), so no value was actually compared 1:1. The paper's central conclusion — polymerase division of labor inferred from ARS leading/lagging ratios, regression and simulations — was deliberately excluded because it relies on the authors' non-public custom code (only third-party Ribose-Map is public). The limitation sits on the authors'/availability side (q4 red) rather than reflecting any measured deviation or fabrication; severity and core-claim status are simply not establishable from the current artifacts.
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-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.