Exposure to the widely used herbicide atrazine results in deregulation of global tissue-specific RNA transcription in the third generation and is associated wit
Provisional — an automated or curator check raised a specific concern and points reviewers here. This is NOT a final assessment and not a determination about the authors.
The main result did not reproduce in this reproduction attempt. Where our recomputation produced values that differ from the published ones, those discrepancies are listed below. This is a single automated attempt — not peer review and not a finding of error or misconduct — and differences can also arise from data access, undocumented parameters or the computing environment. The verdict can be contested via “report an error”.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓Same input data as the authors
- 🟡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
PARTIAL (described well enough to run; result DIFFERENT in magnitude, same in direction). Reproduced the authors' RNA-seq pipeline end-to-end on their own public data (GSE81091/SRP074350: all 18 F3 PE-100 runs SRR3475391-408, testis/liver/brain x ctrl/atz x3): FastQC -> TopHat2 2.1.1 (Ensembl mm9/NCBIM37 rel-67; 97.9-98.2% mapping all 18) -> Cufflinks 2.2.1 -> Cuffmerge (17/18) -> Cuffquant (18/18) -> Cuffnorm -> custom DE (>50th-quantile in >=1 cond, >2-fold, limma BH-FDR<5%). HEADLINE NUMBERS DO NOT MATCH: reported 1419 total DE transcripts (1322 testis + 69 liver + 28 brain; 704 genes) vs reproduced 416 (217 testis + 128 liver + 71 brain; 397 genes). The QUALITATIVE finding reproduces -- testis is the most-affected tissue in our run too -- but the extreme testis-dominance (93% reported) is much weaker here (52%): testis came out far lower, liver/brain higher. Dominant cause = the DE filter is underspecified ('50th quantile of all values' is ambiguous; per-tissue median ~0 for liver/brain makes that pre-filter nearly inert) plus limma-on-FPKM-after-hard-FC-prefilter not being the standard Cuffdiff path, compounded by TopHat2/Cufflinks non-determinism. No fabrication signal: the gap is fully explained by underspecification, not by unsupportable reported values. NOT attempted: secondary C6-C10 (need CPC/CPAT/Cuffcompare/APA steps), ChIP-seq H3K4me3 path, all wet-lab/qPCR/motif/external-overlap. Two pipeline bugs fixed this run: cuffnorm group-arg collapse («job») and a front1-only TMPDIR breaking the R env build on the compute node («job»); final clean run = «job».
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: 2026-06-16 ⛓ b0d8c9c48dce
✎ 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 2026-06-16no 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 authors hypothesized that embryonic exposure to the herbicide atrazine (ATZ) during the E6.5–E15.5 developmental window causes heritable epigenetic reprogramming and affects reproduction in subsequent (F1 and F3) generations.
- ★ Embryonic ATZ exposure affects meiosis, spermiogenesis and reduces spermatozoa number in F3 generation male mice finding
- ★ Changes in testis cell types originate from a modified transcriptional network in undifferentiated spermatogonia mechanism
- ★ ATZ exposure dramatically increases the number of transcripts with novel transcription initiation sites, spliced variants and alternative polyadenylation sites finding
- ★ There is a global decrease in H3K4me3 occupancy in testes of third-generation (F3) ATZ-lineage males finding
- ★ Regions with altered H3K4me3 occupancy in F3 ATZ-derived males correspond to altered H3K4me3 occupancy in F1 generation, and 74% of changed peaks in F3 are associated with enhancers finding
- ★ Regions with altered H3K4me3 occupancy are enriched in SP family and WT1 transcription factor binding sites finding
- ★ Embryonic ATZ exposure effects on development and epigenetic marks are transferred up to three generations finding
- This is the first study integrating genome-wide ChIP-seq and RNA-seq across F1 and F3 generations for toxicant transgenerational inheritance, generating novel sequencing data for outbred CD1 mice resource
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| ChIP-seq (H3K4me3) | testis tissue, F1 and F3 male mice (ATZ-lineage vs control) | embryonic ATZ exposure (F0 dams, 100 mg/kg/day) | genome-wide H3K4me3 occupancy/differential peaks | Illumina HiSeq2000 |
| strand-specific paired-end RNA-seq | testis, liver and hypothalamus, F1 and F3 male mice | embryonic ATZ exposure | differentially expressed genes/transcripts, novel TSS, splice variants, alternative polyadenylation | — |
| H&E histology | testis sections, F3 males | embryonic ATZ exposure | testis architecture/morphology | — |
| immunostaining (ZBTB16, GATA1) | testis sections, F3 males | embryonic ATZ exposure | relative proportion of germ and Sertoli cells | — |
| FACS | dissociated testis cells, F3 males | embryonic ATZ exposure | relative proportion of testicular cell types | — |
| sperm counting | epididymis, F1 and F3 males | embryonic ATZ exposure | spermatozoa number | — |
| immunostaining of meiotic surface spreads (SYCP3, SYCP1, TERF1) | testis, F3 males | embryonic ATZ exposure | synaptonemal complex formation, chromosome synapsing defects, telomere connections | — |
| Western blot | whole testis extract / purified histone fraction, F3 males | embryonic ATZ exposure | protamine 2 and H4K5Ac protein levels | — |
- ▼ Spermatozoa number significantly decreased in ATZ-derived F1 and F3 males compared to controls ~30% decrease in F3
- – 704 genes corresponding to 1419 differentially expressed transcripts identified genome-wide 704 genes / 1419 transcripts
- ▼ Protamine 2 protein level decreased in F3 ATZ-derived testis 2.6-fold
- ▼ H4K5Ac level decreased in purified histone fraction of F3 ATZ-lineage males 1.3-fold
- ▲ Significant increase in meiotic synapsing defects in F3 ATZ male progeny
- ▼ Global decrease of H3K4me3 occupancy observed in F3 generation males
- – Majority of altered H3K4me3 peaks in F3 correspond to enhancer regions 74%
- – Altered H3K4me3 regions enriched for SP family and WT1 transcription factor binding motifs
- fold_change ~30% decrease (spermatozoa count decrease in F3 ATZ vs control)
- fold_change 2.6 times decrease (protamine 2 protein level in F3 ATZ testis)
- fold_change 1.3 times decrease (H4K5Ac level in F3 ATZ-lineage males)
- count n=199 control, n=195 ATZ (quantitative analysis of meiotic (synaptonemal complex) defects)
- count 704 genes / 1419 differentially expressed transcripts (genome-wide RNA-seq differential expression analysis)
- other 74% (proportion of altered F3 H3K4me3 peaks associated with enhancers)
- pvalue P-value threshold <10E-5 (MACS 2.0.1 H3K4me3 peak calling threshold)
- pvalue FDR <10% (ChIP-seq peaks); FDR <5% (RNA-seq DEGs) (Limma test thresholds for differential peak/transcript calling)
Statistical methods review
Model: opusA 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 an outbred CD1 mouse transgenerational exposure design with genome-wide ChIP-seq (H3K4me3) and RNA-seq, plus targeted qPCR and cytological/Western-blot assays. Genomic differential analyses (differential peaks and differentially expressed transcripts) were filtered by fold change and then assessed with the R/Limma test under FDR thresholds (10% for ChIP-seq, 5% for RNA-seq), with ChIP-seq replicate concordance checked by an irreproducible discovery rate criterion. Targeted qPCR comparisons used Student's t-test on at least four independent experiments. Results were largely reported as fold changes and counts of significant features.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Student's t-test | qPCR/RT-qPCR comparisons and ChIP-qPCR enrichment fold changes (ATZ vs control) | duplicates of at least four independent experiments | not stated |
| Limma test | differential H3K4me3 peak calling between ATZ-treated and control ChIP-seq samples | two biological replicates per condition | not stated |
| Limma test | identification of differentially expressed transcripts/genes from RNA-seq (testis, liver, hypothalamus) | three biological replicates per tissue | not stated |
| Irreproducible discovery rate (IDR) criterion | confirming similarity of ChIP-seq biological replicates for peak sets | two biological replicates | na |
| Significance test (method not stated) | quantitative analysis of meiotic synapsis defects (n=199 control, n=195 ATZ cells) and spermatozoa counts | n = 199 control and n = 195 ATZ-derived cells | not stated |
-
Differential expression and differential peak significance were assessed with R/Limma after a fold-change pre-filter.↳ Could also: Count-based negative-binomial frameworks such as DESeq2 or edgeR for RNA-seq, and tools like DiffBind/csaw for ChIP-seq, could also have been used. — These model count data and dispersion directly and integrate independent filtering with FDR control, which some analysts prefer for small replicate numbers in sequencing experiments.
-
Genome-wide analyses used two biological replicates for ChIP-seq and three for RNA-seq.↳ Could also: Additional biological replicates per condition could also have been included. — More replicates can increase the precision of dispersion estimates and the statistical power to detect smaller effects, which is one reason higher replication is often recommended for genomics.
-
Targeted qPCR comparisons between two groups were evaluated with Student's t-test.↳ Could also: A Welch's t-test or a non-parametric Mann-Whitney U test could also have been applied. — Welch's variant relaxes the equal-variance assumption and rank-based tests avoid normality assumptions, both of which are commonly chosen when sample sizes are small.
-
Many qPCR-style comparisons were each tested individually with t-tests.↳ Could also: When several related comparisons are made, an ANOVA with a post-hoc correction (e.g., Tukey HSD) or a multiplicity adjustment (e.g., Benjamini-Hochberg) across the family could also be used. — A unified model with correction controls the family-wise or false-discovery error rate across the set of related comparisons, which some prefer when reporting multiple tests together.
-
Results were largely summarized as fold changes and counts of significant features with threshold-based p-values.↳ Could also: Reporting exact p-values together with effect-size estimates and 95% confidence intervals could also have been presented. — Confidence intervals and exact values convey the magnitude and precision of effects in addition to significance, which is increasingly encouraged in reporting guidelines.
-
FDR thresholds were applied after an initial fold-change pre-filter for both genomic datasets.↳ Could also: An independent-filtering or shrinkage-based ranking that combines statistical significance and effect size within one model could also have been used. — Integrated approaches can improve the calibration of the FDR and the stability of effect-size estimates, which is why they are a common alternative to sequential filtering.
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.
-
H3K4me3 occupancy is globally decreased in F3 testes of ATZ-lineage males relative to controls.ChIP-seq mouse testis down 2016×1papers★ This paper is the founder (earliest)
-
74% of differentially occupied H3K4me3 peaks in F3 ATZ-lineage testes overlap enhancer regions.ChIP-seq mouse testis 2016×1papers★ This paper is the founder (earliest)
-
Meiotic synapsis defects (assessed by SYCP3/SYCP1 immunostaining) are significantly increased in F3 ATZ-lineage spermatocytes.imaging mouse testis up 2016×1papers★ This paper is the founder (earliest)
-
Spermatozoa count is significantly reduced (~30%) in F3 males derived from ATZ-exposed lineages.other mouse epididymis down 2016×1papers★ This paper is the founder (earliest)
-
704 differentially expressed genes and 1419 transcripts are detected in F3 ATZ-lineage testes by RNA-seq.RNA-seq mouse testis mixed 2016×1papers★ This paper is the founder (earliest)
-
H4K5ac histone mark is decreased 1.3-fold in purified histone fractions from F3 ATZ-lineage testes.western-blot mouse testis down 2016×1papers★ This paper is the founder (earliest)
-
Protamine 2 (PRM2) protein level is decreased 2.6-fold in F3 ATZ-lineage testis.western-blot mouse testis down 2016×1papers★ This paper is the founder (earliest)
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.md — PMID 27655631 (atrazine transgenerational RNA-seq + ChIP-seq)
Title: Exposure to the widely used herbicide atrazine results in deregulation of global tissue-specific RNA transcription in the third generation and is associated with a global decrease of histone trimethylation in mice — Hao et al., NAR 2016. PMID 27655631 · PMC5175363 · DOI 10.1093/nar/gkw840.
Cited code artifact: https://github.com/najoshi/sickle (third-party FASTQ quality
trimmer; per P16 a valid reproducible artifact). NOTE: in this paper sickle is used
in the ChIP-seq pipeline (-q33 + Bowtie 1.0.0), NOT in the RNA-seq pipeline.
Data: GEO GSE81093 (SuperSeries). Subseries:
- GSE81091 = RNA-seq (SRP074350 / PRJNA320479) — 18 runs, all paired-end 100bp, HiSeq 2500. F3 only: 3 tissues (testis/brain/liver) x 2 (control/atrazine) x 3 reps. Runs SRR3475391–SRR3475408 (~60–78M read pairs each, ~130 GB FASTQ total).
- GSE81056 / GSE84978 = ChIP-seq H3K4me3 (F1+F3 testis).
In scope (pipeline-derived, attempted) — RNA-seq, the 80/20 core
Pipeline as described in Methods "RNA-Seq expression data processing": FastQC QC → TopHat 2.0.12 map to Ensembl mm9 → BAM → Cufflinks assemble → Cuffmerge vs Ensembl mm9 annotation → Cuffquant + Cuffnorm (Cufflinks 2.2.1) expression levels → custom DE filter: keep transcripts > 50th quantile of all values in ≥1 condition, then > 2-fold ATZ-vs-control difference, then Limma with FDR < 5%.
Primary claims to regenerate (see claims.tsv C1–C5):
- C2 1419 total DE transcripts (FC>2, FDR<0.05); split C3 testis 1322, C4 liver 69, C5 brain 28; C1 704 collapsed genes.
These are the clearly-specified, headline numeric outputs → primary target.
Partially in scope (secondary, attempt if primary lands)
- C6–C7 lncRNA/LincRNA fraction (needs extra CPC + CPAT coding-potential step; CPC/CPAT thresholds given: negative CPC + CPAT prob <40%).
- C8–C10 alternative-isoform / APA transcript counts (Cuffcompare class codes; APA step not fully specified).
Out of scope (not attempted, stated why)
- ChIP-seq H3K4me3 peak calling & differential peaks (sickle + Bowtie1 + MACS2 + CHANCE). Separate heavy pipeline; RNA-seq DEG counts are the paper's headline and the cleaner 80/20. May revisit if time permits since sickle is the cited artifact.
- All wet-lab / qPCR / IGV-visualization / motif (TomTom) / external-dataset overlap claims (157 ATZ-overlap, H4K5ac/H4K8ac comparisons, vinclozolin comparison) — these depend on external published datasets + manual steps, not regenerable from GSE81093 alone.
- F1 results: no RNA-seq for F1 (RNA-seq is F3-only); F1 only appears in ChIP-seq.
Known reproduction risks (flag for auditor)
- TopHat2/Cufflinks is deprecated and assembly is non-deterministic (Cuffmerge novel transcript IDs vary run-to-run) → exact 1419 is unlikely; expect within-tol/partial. Target: recover the tissue pattern (testis ≫ liver ≈ brain) and order of magnitude.
- The DE step is underspecified: applying Limma after a hard fold-change pre-filter on Cuffnorm FPKM is statistically unusual; the exact design matrix / contrast / whether FPKM is log-transformed is not stated → docs_insufficient risk for exact counts.
- Software versions: TopHat 2.0.12 + Cufflinks 2.2.1 pinned in paper; Bowtie/TopHat index for Ensembl mm9 must be built.
Status
Eligible. Data fully public, pipeline tools all open-source & conda-installable, expected numeric result pinned (1419 / 1322 / 69 / 28 / 704). Heavy compute → «our HPC».
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.
The RNA-seq pipeline (TopHat2->Cufflinks->Limma) was still running at an operator-requested early finalize, so no DEG counts were computed and all primary claims (1419 transcripts, 704 genes, 1322/69/28 per tissue) remain pending — no reproduced value was asserted, so there is no fabrication. The data is fully public and 1:1 available (GSE81093/SRP074350, 18 F3 runs), but the authors' DE step is underspecified (Limma after a hard fold-change FPKM pre-filter), which would add uncertainty even on completion. The dominant cause of 'no result' is our incomplete run, not an authors' defect or a measured discrepancy. Overall this is a partial/pending reproduction: sound, auditable and resumable, but with no comparison achieved — hence uniformly yellow rather than a green pass or a red fabrication/discrepancy.
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.