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

Analysis of whole-genome re-sequencing data of ducks reveals a diverse demographic history and extensive gene flow between Southeast/South Asian and Chinese pop

Genet Sel Evol · 2021
L1 23/100 3/4
⚑ Flagged for review — a reproduced result did not match the reported value

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.

Why this verdict

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”.

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: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
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 +6
✓ What held up
  • Reported values were directly comparable
What did not (or only partly)
  • 🟡Could not use the authors’ exact input data
  • 🟡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
23/100
Reproducibility score
2.9 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 1% of all assessed papers rank 1160 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

PARTIAL reproduction (1:1 on pipeline + magnitude + significance; directional mismatch on the 2 high-power quartets). Reproduced all 3 Table-S5 ABBA-BABA D-statistics end-to-end from raw CRA002628 FASTQ via the authors' own abba_baba tool (github.com/owensgl/abba_baba @993928a) on a 20-duck quartet subset: BWA-MEM -> GATK HaplotypeCaller -> 16-way parallel joint genotyping -> paper hard-filter -> 559,718 SNPs -> 2 Mb block-jackknife D. We recover SIGNIFICANT SE-Asian<->South-Asian gene-flow signals of the right order of magnitude (cambodia |D|=0.0267 vs reported 0.0247), but the SIGN is reversed for Vietnam (D=-0.043 vs +0.018) and Cambodia (D=-0.027 vs +0.025) -- attributing the excess allele-sharing to Pakistan rather than Bangladesh -- while Laos matches in sign (+0.157 vs +0.065) but is underpowered (433 sites). Best explanation: the reduced/different P1,P2 sample sets (2 Bangladesh, 3 Pakistan vs paper 4, 5). Labels + P1/P2/P3 convention independently verified vs Table S1/S5. Both datasets profiled (CRA002628 grade B subset-verified, SRP144280 grade A metadata). NOT attempted: full 109-duck panel; PSMC/ADMIXTURE/PCA/trees (other tools, out of scope).

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 23
    assessed: 2026-06-21 ⛓ 126c668133eb
✎ 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
2026-06-21
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-21
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: sonnet
Founding hypothesis

The study tests what the demographic history, gene flow, and domestication patterns are between Southeast/South Asian and Chinese duck populations, given that little was previously known about domestication in Southeast/South Asian ducks.

Core claims
  • Whole-genome resequencing reveals three geographically distinct genetic groups: local Chinese, wild, and local Southeast/South Asian duck populations finding
  • Chinese domestic ducks experienced the strongest population bottleneck, caused by both domestication and the last glacial maximum, while Chinese wild ducks experienced a weaker bottleneck from domestication only finding
  • The bottleneck was more severe in Southeast/South Asian populations than in local Chinese populations, yielding a smaller effective population size (7100-11,900) for the former finding
  • Extensive gene flow occurred between Southeast/South Asian and Chinese populations, and between Southeast Asian and South Asian populations, with prolonged gene flow between Guangxi (China) and neighboring SE/S Asian populations finding
  • A genomic region on duck chromosome 1 containing PNPLA8, THAP5, and DNAJB9 shows high probability of gene flow between Guangxi and Southeast/South Asian populations finding
  • Strong selection signatures were detected in genes involved in nervous system development signaling (e.g., ADCYAP1R1, PDC) and morphological traits such as cell growth (e.g., IGF1R) finding
  • 109 ducks (78 newly sequenced plus 31 published) were used to generate what is described as the most comprehensive catalog of genetic variants in ducks (4,054,630 high-quality SNPs) resource
  • Southeast Asian and South Asian populations are sister groups, as are the LC x BY duck lineages from Fujian relative to other Chinese domestic populations finding
Experimental setups
Assay System Perturbation Readout Platform
whole-genome resequencing (Illumina HiSeq 2500, ~5x coverage) 78 newly sequenced ducks (domestic, wild, muscovy, Southeast Asian, South Asian) plus 31 published sequences (109 total individuals) none SNP/indel genotypes across the genome Illumina HiSeq 2500
phylogenetic and PCA analysis (SNPhylo, GCTA) 109 duck individuals across 8 domestic, 2 wild, 2 Muscovy, 3 SE Asian, 2 S Asian populations none population clustering, phylogenetic tree topology, principal components SNPhylo v20140701; GCTA v1.91.6
population structure/admixture analysis (Admixture, CLUMPAK) 109 duck individuals, K=2 to 9 clusters none ancestry proportions per individual Admixture v1.3.0; CLUMPAK v1.1
effective population size and divergence time inference (SMC++) duck populations (wild, domestic, SE/S Asian) none historical effective population size (Ne), divergence time between wild and domestic populations SMC++ v1.15.2
admixture graph modeling (TreeMix) and three-population test (f3-statistic) Southeast/South Asian, Chinese indigenous, and wild duck populations none migration edges/proportions, evidence of admixture TreeMix v1.13
D-statistic (ABBA-BABA) introgression test and fd sliding-window scan LC (P1), Chinese indigenous (P2), Southeast/South Asian (P3), C. moschata (outgroup) none D-statistic, Z-scores, fd values across 100-kb windows (20-kb steps) custom Perl script (github.com/owensgl/abba_baba)
phased haplotype coancestry inference (ChromoPainter/FineStructure, Globetrotter) phased SNP data from all duck individuals (Shapeit2-phased) none coancestry matrices, population clustering, ancestral donor proportions ChromoPainter v2; FineStructure v4.0.1; Shapeit2
genome scan for selection signatures (theta-pi, FST, Tajima's D) 44 domestic ducks (25 domestic + 19 highly selected) vs 19 wild ducks, 40-kb sliding windows (10-kb steps) domestication/artificial selection theta-pi ratio, Z(FST), Tajima's D, KEGG-enriched candidate genes (1049 genes) VCFtools v0.1.15
Key results
  • 3,902,414 high-quality SNPs used to cluster 109 ducks into three major genetic groups via NJ tree and PCA 3,902,414 SNPs
  • PC1 and PC2 explained variance in population structure PC1=8.66%, PC2=4.15%
  • FST values between Guangxi and Southeast/South Asian populations were close to those between Guangxi and Chinese indigenous populations 0.024-0.115 vs 0.052-0.135
  • Southeast/South Asian populations showed smaller effective population size than local Chinese populations due to more severe bottleneck Ne = 7100-11,900
  • Prior literature estimate: divergence between Chinese domesticated and wild ducks occurred ~2200 generations ago
  • Prior literature estimate: gene flow (migrants per generation) from meat and egg/dual-purpose breeds into wild duck 1.12 and 3.92 migrants/generation
Key statistics
  • count 4,054,630 high-quality SNPs (genome-wide SNP set after PLINK filtering (-geno 0.05 -maf 0.05) used for phylogenetic/population genetic analyses)
  • count 3,902,414 high-quality SNPs (SNP set used for NJ tree and PCA after quality control)
  • other effective population size 7100-11,900 (SE/S Asian populations following bottleneck)
  • fold_change log2(theta-pi ratio) = 0.69 (top 5% threshold) (threshold for significant selection signature based on theta-pi wild/domestic ratio)
  • other Z(FST) = 1.86 (top 5% threshold) (threshold for significant selection signature based on FST)
  • other absolute Z-scores > 3 (threshold for strong evidence of admixture in D-statistic block jackknife test)
  • count 1049 genes (genes within 40-kb regions with significant selection signatures, submitted to KEGG enrichment analysis)
  • other FST 0.024-0.115 (Guangxi vs SE/S Asian) and 0.052-0.135 (Guangxi vs Chinese indigenous) (pairwise population differentiation comparison)

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.

This is a population-genomics study, not an experimental biomedical trial: whole-genome resequencing data from 109 ducks were analyzed using phylogenetic (neighbor-joining/maximum-likelihood tree), clustering (PCA, ADMIXTURE, fineSTRUCTURE), demographic (SMC++, ABC-GLM), and admixture/gene-flow (TreeMix f3, D-statistic/ABBA-BABA, fd) methods. Genome-wide selection scans used sliding windows with empirical outlier thresholds (top 5%) on θπ ratios and FST rather than classical hypothesis-testing p-values. Uncertainty was expressed mainly through block-jackknife standard errors/Z-scores, bootstrap support, and replicate analyses rather than through conventional group-comparison statistics (t-tests, ANOVA).

Replicationbiological Sample sizeExplicit per-population sample sizes given (e.g., 78 newly sequenced + 31 published = 109 individuals; 44 domestic vs 19 wild ducks for the selection scan); no formal power calculation described GroupsGeographically/genetically defined duck populations (Chinese domestic, Chinese wild, Southeast Asian, South Asian, Muscovy outgroup) Pairingna Randomization/blindingnot stated Dispersionmixed Effect sizesyes Multiplicity correctionNone stated; genome-wide scans used empirical outlier thresholds (top 5% for θπ ratio and FST) and a fixed |Z|>3 cutoff for D-statistics rather than a formal multiple-testing correction
Statistical tests used
Test Applied to n Assumptions
Three-population test (f3 statistic, TreeMix) Admixture graph / test for admixture among Southeast/South Asian, Chinese indigenous, and wild populations 109 individuals; standard errors from blocks of 500 SNPs not stated
D-statistic (ABBA-BABA), block-jackknife Introgression test between Chinese (P1/P2, e.g. LC) and Southeast/South Asian (P3) populations, with C. moschata as outgroup genome-wide SNPs; jackknife block size 2 Mb; |Z|>3 used as evidence threshold not stated
fd statistic, sliding-window scan Localizing gene flow between Guangxi and Southeast/South Asian populations 100-kb windows, 20-kb steps, minimum 100 SNPs/window not stated
FST and dxy divergence statistics Confirming the top introgressed region between Guangxi and Southeast/South Asian ducks; also population pairwise differentiation (Fig. 1) genome-wide SNPs per population pair not stated
Genome scan: θπ ratio, Z-transformed FST, Tajima's D with top-5% empirical outlier cutoff Detecting selection signatures between 44 domestic and 19 wild ducks 40-kb windows, 10-kb steps, 63 individuals total not stated
ABC-GLM model comparison (ABCtoolbox) Evaluating alternative demographic/simulation models not stated in detail (referred to Appendix) not stated
Approaches that could also have been used
  • The genome-wide selection scan (θπ ratio, FST) used a fixed top-5% empirical outlier threshold to flag candidate regions
    Could also: A coalescent-simulation-derived null distribution (e.g., via ms/msprime) or a formal false-discovery-rate procedure (e.g., Benjamini-Hochberg) applied across all genomic windows — This would assign explicit significance levels/p-values to outlier windows and quantify the expected false-positive rate across the many windows tested, complementing the descriptive percentile cutoff
  • Evidence for admixture/introgression from the D-statistic was judged using a fixed |Z|>3 threshold from block-jackknifing
    Could also: Converting jackknife Z-scores to two-sided p-values and applying a multiple-testing correction (e.g., Bonferroni) across the full set of population-trio comparisons tested — This would make the significance criterion continuous and explicitly adjust for the number of D-statistic tests performed, which can be informative when many population trios are compared simultaneously
  • Effective population size history was inferred with SMC++ without phasing
    Could also: Complementary LD-based (e.g., SNeP) or haplotype-based (e.g., PSMC, MSMC2) methods for estimating historical Ne — Cross-validating recent demographic history estimates across methods with different assumptions can help characterize how estimates depend on modeling choices, especially for very recent time scales
  • Demographic/admixture models were compared using an ABC-GLM approach in ABCtoolbox
    Could also: Full-likelihood composite-likelihood methods such as fastsimcoal2 or ∂a∂i for demographic model fitting — These approaches can jointly estimate model parameters with associated confidence intervals in addition to selecting among candidate models, which some readers may find complementary to an ABC model-choice framework
  • Population relationships were summarized with a neighbor-joining/maximum-likelihood tree (SNPhylo) supported by 1000 bootstraps
    Could also: An explicit admixture-graph or network method (e.g., the already-used TreeMix, or alternatives like OrientAGraph) as the primary relationship depiction rather than a strictly bifurcating tree — Because gene flow is a central finding of the paper, a network-based representation can jointly display both splits and admixture edges in one diagram, which some readers find more directly interpretable alongside a bifurcating tree
  • Population structure was characterized with ADMIXTURE across a pre-defined K range (2-9)
    Could also: Reporting cross-validation error curves to formally justify the chosen K, alongside complementary discriminant methods such as DAPC — This provides an additional quantitative criterion for selecting the most supported number of ancestral clusters, complementing visual inspection of the ADMIXTURE plots
Software: Trimmomatic 0.36 · BWA-MEM 0.7.15 · Samtools 1.3.1 · Picard MarkDuplicates 1.108 · GATK 3.5 · PLINK 1.9 · ADMIXTURE 1.3.0 · CLUMPAK 1.1 · SNPhylo 20140701 · GCTA 1.91.6 · SMC++ 1.15.2 · ABCtoolbox 2.0 · TreeMix 1.13 · Shapeit2 · ChromoPainter/FineStructure 4.0.1 (FineStructure) · PopLDdecay 3.40 · VCFtools 0.1.15

What was reproduced

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

scope.md — pmid-33849442 (duck WGS demographic history & gene flow)

Paper: Analysis of whole-genome re-sequencing data of ducks ... gene flow between Southeast/South Asian and Chinese populations. Genet Sel Evol 2021. DOI 10.1186/s12711-021-00627-0 · PMCID PMC8042899.

Code (P16 — third-party generic tool)

github.com/owensgl/abba_baba — Greg Owens' GENERIC ABBA-BABA D-statistic Perl/R pipeline (hardcoded to sunflower Exil.GATK.2015.tab in the wrapper .sh; the duck authors applied this tool to their own data). Components:

  • ABBA_BABA.v1.pl : per-SNP ABBA/BABA, D numerator/denominator, fhom, fd. Input = a .tab genotype table (chr, pos, per-sample IUPAC or 2-letter genotypes) + poplist (sample\tpop) + group file (pop\t[1-4] = P1,P2,P3,O).
  • ABBA_out_blocker.pl : sums Num/Denom into genomic blocks (window_size hardcoded).
  • Jacknife_ABBA_pipe.R : block-jackknife std error (needs R bootstrap).
  • ABBA_pvalue.R : Z = D/SE, p = 2*pnorm(-|Z|).
  • MISSING DEP: require "countbadcolumns.pl" (from «path», NOT in repo) — auto-detects IUPAC coding + number of leading metadata columns. Must reconstruct.

Data (TWO datasets — registry only lists one)

  • SRP144280 (SRA/ENA, OPEN): 106 WGS runs Anas platyrhynchos, ~485 GB, 1.47 Tbp. Paper reuses 31 of these (21 Chinese indigenous + 10 Chinese wild).
  • CRA002628 (GSA China ngdc.cncb.ac.cn, OPEN, HTTPS/FTP): 78 items / 156 files / 940 GB. The 78 NEWLY sequenced ducks incl. the KEY samples for the D-stats: P1=LC(Liancheng), SE/S-Asian P3 (Vietnam/Cambodia/Laos/Pakistan/Bangladesh), muscovy (Cairina moschata) outgroup. NOT in the registry — without it NO reported D-statistic is reproducible.

Pipeline reported (Methods)

raw FASTQ -> BWA-MEM v0.7.15 (-t -M -R) -> GATK HaplotypeCaller (genotype likelihoods) -> GATK hard filter (QUAL>30, QD>2, FS<60, MQ>40, MQRankSum>-12.5, ReadPosRankSum>-8, SOR>10) -> PLINK --geno 0.05 --maf 0.05 -> 4,054,630 SNPs -> abba_baba Perl (D-stat, 2 Mb non-overlapping block jackknife, Z>3 = admixture). Reference: IASCAAS_PekingDuck_PBH1.5.

In scope (pipeline-derived)

D-statistics (Table S5) for the (((P1,P2),P3),O) quartets:

  • Guangxi P2: D 0.033–0.054, Z>3
  • Anhui/Guizhou vs Laos: D 0.0179 / 0.0173, Z>3
  • Pakistan(P1) vs Bangladesh(P2) vs SE-Asia(P3): D 0.0182–0.0645, Z>3 Reproduce via the authors' abba_baba tool on a genotype table built from the data.

Out of scope

Wet-lab sequencing, PSMC/demographic history curves, ADMIXTURE/PCA structure, phylogenetic trees (different tools) — unless time permits after the D-stat core.

Feasibility / plan

Full reproduction = align 109 ducks (~1.4 TB FASTQ) + GATK joint genotyping -> 4M SNPs -> abba_baba. Very heavy (multi-week HPC). Pragmatic FLOOR target: a FOCUSED end-to-end reproduction of ONE reported quartet using the actual abba_baba tool — download only the ~15–25 samples of one quartet (P1,P2,P3,O), BWA+GATK call them jointly, build the .tab table, run the Perl, compare D & Z to Table S5. Plus full DATASET PROFILING of both accessions (cheap, high value). Honest partial is the expected outcome; document scale truthfully.

Figures / tables: Table
D_PAK_BAN_VIE
Reported
D=0.018200 sd=0.004370 Z=4.165
Reproduced
D=-0.043038 sd=0.010254 Z=-4.197 (43121 sites)
did not match
D_PAK_BAN_CAM
Reported
D=0.024734 sd=0.004290 Z=5.765
Reproduced
D=-0.026657 sd=0.010783 Z=-2.472 (40531 sites)
did not match
D_PAK_BAN_LAO
Reported
D=0.064541 sd=0.004895 Z=13.186
Reproduced
D=+0.156777 sd=0.060200 Z=2.604 (433 sites)
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 23/100

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: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
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 +6

End-to-end reproduction via the authors' own abba_baba tool on a 20-duck subset (vs 109) recovers significant gene-flow signals at the right order of magnitude (Cambodia |D|=0.0267 vs 0.0247), but the direction reverses for both well-powered quartets (Vietnam +0.0182→-0.043; Cambodia +0.0247→-0.027) and Laos is sign-correct yet underpowered (433 sites, |D| 2.4x). The deviation sits on the input/sample-definition side and is our own methodology (reduced, partly-failed P1/P2 sample sets), not an authors' defect — reported values are internally consistent with PMC text and show no fabrication indicator. Factually the sign flip is severe (q6 red), but the broad gene-flow conclusion survives as a significant signal, so overall this is a solid reproduction with explainable deviations pending a full-panel rerun.

🤝
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.

2.9 M
tokens (I/O) · 305.3 M incl. cache
1344 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.