Whole-genome sequencing of cryopreserved resources from French Large White pigs at two distinct sampling times reveals strong signatures of convergent and diver
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.
- ✓Same input data as the authors
- ✓Reported values were directly comparable
- ✓No relevant deviation in data/preprocessing
- ✓No authors-side cause for any deviation
- ✓Reported values are derivable from the shared data
- ✓Any deviation was negligible
- ✓The central claim held under reproduction
- ✓Overall, the reproduction was clean
- Every checked point held up.
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; clean 1:1 reproduction of the genetic-diversity layer. Ran the authors' own pipeline (PLINK 1.90b6.21 --freq --family + scripts/snp_stat.R, repo commit 169cd7e) on the authors' Zenodo HQSNP genotypes (10.5281/zenodo.6415023, snp20_auto_cr, 3 files SHA256-verified) on «our HPC» SLURM «job». All 8 in-scope deterministic claims matched the paper to the reported precision: HQSNP count 13,408,342 (exact integer), 36-animal composition (exact), and all six per-population polymorphic/private-allele fractions (83.1/79.2/76.0% and 6.9/5.3/4.3% -> reproduced 83.13/79.17/76.04% and 6.93/5.26/4.27%). No fabrication concern: every reported value is directly derivable from the shipped Zenodo data via the shipped script. NOT attempted (80/20, see scope.md): Ne estimates (stochastic 100k-SNP thinning, no seed pinned in repo -> not byte-reproducible), genome-wide selection signatures (hapFLK + temporal-HMM over 13.4M SNPs, Table 1 / 151 genes -- the heavy multi-tool ~20%), and variant calling from raw PRJEB51909 reads (3-caller consensus front end). Note: registry data_accession GSE56011 is the paper's gene-expression GEO series, not the WGS genotypes -- the reproducible genotypes live on Zenodo 6415023.
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 100assessed: 2026-06-15 ⛓ eed7bba22c96
✎ 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 2026-06-15👤 1 human curator(s) · Level L2 2026-06-15
- 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: sonnetBy sequencing genomes from cryopreserved samples taken at multiple time points (1977 ancestral population versus modern LWD and LWS lines diverged since 1995), the study tests whether it is possible to detect and finely characterize signatures of recent selection—distinguishing convergent, divergent, and line-specific selection events and their underlying traits/genes—in French Large White pigs.
- ★ French LWD and LWS lines have lost approximately 5% of the SNPs that segregated in the 1977 ancestral population. finding
- ★ 38 genomic regions under recent selection were detected and classified as convergent (18), divergent (10), dam-line-specific (6), or sire-line-specific (4). finding
- ★ Enriched biological functions differ by category: body size/weight/growth across all region categories, early life survival and calcium metabolism specifically in dam-line signatures, and lipid and glycogen metabolism specifically in sire-line signatures. finding
- ★ Recent selection on IGF2 was confirmed, and several other regions were linked to single candidate genes (ARHGAP10, BMPR1B, GNA14, KATNA1, LPIN1, PKP1, PTH, SEMA3E, ZC3HAV1, among others). finding
- ★ A combined methodology (hapFLK differentiation scan plus a temporal HMM-based allele-frequency method with a local score approach) was used to detect and classify candidate selection regions. method
- ★ Sequencing genomes of animals at several recent time points from cryobank resources generates considerable insight into traits, genes and variants under recent selection, and is applicable to other livestock populations. finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Whole-genome sequencing (WGS) | LW boars born 1977 (n=10), blood/semen | none | SNP/indel genotypes, allele frequencies | Illumina HiSeq X Ten, 2x150bp, NEBNext Ultra DNA Library Prep Kit |
| Whole-genome sequencing (WGS) | LWD boars 2014-2015 (n=13), blood | none (recent selection line) | SNP/indel genotypes, allele frequencies | Illumina HiSeq X Ten |
| Whole-genome sequencing (WGS) | LWS boars 2012-2016 (n=13), blood | none (recent selection line) | SNP/indel genotypes, allele frequencies | Illumina HiSeq X Ten |
| Genetic diversity analysis (MDS on IBS distance matrix) | 36 pig genomes (1977, LWD, LWS) | none | population structure, total/private polymorphism counts, effective population size | PLINK 1.9; R 3.5.1 cmdscale(); R package NB |
| Genome-wide selection scan (FLK/hapFLK statistic) | LWD vs LWS modern lines (HQSNP set) | none (comparative population genetics) | haplotype/allele frequency differentiation, q-values | hapFLK software v1.4 |
| Temporal allele-frequency selection test (HMM method of Foll et al. + local score) | 1977 vs LWD; 1977 vs LWS | none | p-value of selection evidence, selective advantage estimate | custom HMM implementation; local score approach |
| Functional enrichment analysis | Genes in candidate selection regions (and flanking genes) | none | enriched GOBP, KEGG and MGI terms | Genecodis4 web tool |
| Variant functional annotation | Autosomal variant (AV) set within candidate regions | none | functional impact classification (HIGH/LOW/MODERATE/MODIFIER) | SnpEff v4.3t |
- ▼ LWD and LWS lines lost SNPs segregating in the 1977 ancestral population ~5%
- – 38 candidate regions under recent selection detected across four categories 18 convergent, 10 divergent, 6 LWD-specific, 4 LWS-specific
- – Body size, body weight and growth functions enriched regardless of region category
- – Early life survival and calcium metabolism functions enriched more specifically in dam-line (LWD) signatures
- – Lipid and glycogen metabolism functions enriched more specifically in sire-line (LWS) signatures
- – IGF2 confirmed as under recent selection; other regions linked to single candidate genes
- – In 2015, LWD and LWS showed major phenotypic differences reflecting differing selection objectives >3 more piglets/litter, ~14 more days to reach 100 kg, 1.5 mm more back fat in LWD vs LWS
- – hapFLK significance called at FDR (q-value) threshold to characterize convergent/divergent/line-specific selection scenarios q < 0.2
- other ~5% SNP loss (SNPs lost in LWD/LWS relative to 1977 ancestral population)
- count 38 candidate regions (total regions under recent selection detected)
- count 18 convergent, 10 divergent, 6 dam-specific, 4 sire-specific (breakdown of selection region categories)
- count n=36 sequenced animals (10 from 1977, 13 LWD, 13 LWS) (whole-genome sequencing sample sizes)
- other 10x (1977 boars) and 15x (recent boars) (expected sequencing depth)
- pvalue q < 0.2 (FDR 20%) (threshold for significant hapFLK values)
- other chromosome-wide FPR of 1% (local score (LS) detection threshold)
- other >3 more piglets per litter; ~14 more days to 100 kg; 1.5 mm more back fat (2015 phenotypic differences between LWD and LWS lines)
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 study used whole-genome sequencing of 36 French Large White pigs from three cryopreserved samples (1977 ancestral population, modern dam line LWD, modern sire line LWS) to characterize genetic diversity and detect signatures of recent positive selection. The primary genome-wide selection scan combined the hapFLK haplotype-differentiation statistic (FDR controlled at 20% via q-values) with a hidden Markov model-based temporal analysis of allele frequency change between 1977 and each modern line, with regional aggregation via the local score approach (chromosome-wide FPR = 1%). Candidate regions were classified into convergent, divergent, or line-specific selection categories based on cross-test patterns, and functional enrichment of associated genes was assessed using Genecodis4 with FDR correction across three annotation databases.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| hapFLK (haplotype-based extension of FLK/FST, population-tree-corrected) | Genome-wide scan for positive selection signatures based on haplotype differentiation between LWD and LWS, using 1977 population as outgroup | 36 animals (10 LW-1977, 13 LWD, 13 LWS) | not stated |
| FLK (FST-based statistic corrected for population size differences and hierarchical structure) | Characterization of candidate regions (applied to all-variants set within each region in a second pass) | 36 animals (10 LW-1977, 13 LWD, 13 LWS) | not stated |
| HMM-based temporal allele frequency test (method of referenced study [24]) | Detection of allele frequency changes between 1977 population and each modern line separately (time-LWD and time-LWS analyses); provides per-SNP p-value and estimated selective advantage | 10 LW-1977 + 13 LWD or 13 LWS per pairwise comparison | not stated |
| Local Score (LS) approach for regional aggregation of SNP-level p-values | Identification of genomic regions with a local excess of low p-values from temporal analysis; significance threshold set to achieve chromosome-wide FPR of 1% | null | not stated |
| Multidimensional scaling (MDS) of identity-by-state (IBS) pairwise distance matrix | Visualization of genetic diversity and population structure across all three samples | 36 animals | na |
| Moment-based effective population size estimation (method of referenced study [17], NB R package) | Estimation of Ne for each modern line relative to the 1977 population based on allele frequency change at SNPs with MAF > 10% | null | not stated |
| FDR-corrected gene set enrichment analysis (Genecodis4; FDR method of referenced study [34]) | Functional enrichment of genes in candidate selection regions across GOBP, KEGG, and MGI databases using Homo sapiens as reference organism | null | not stated |
| Contingency table residuals (weighted, via wtable() in R) | Quantification of association between QTL trait categories (Both, Both.Sire, Dam) and selection region categories (conv, div, LWD, LWS) | null | na |
-
The hapFLK genome-wide scan used a q-value threshold of 0.2 (20% FDR) to call significant regions↳ Could also: A more stringent FDR threshold (e.g., 5% or 10%) or a permutation-based genome-wide significance threshold could also be applied — Stricter thresholds would reduce the expected proportion of false positives among reported candidate regions; with only 36 sequenced individuals the statistical power is limited, so the 20% FDR reflects a deliberate recall-versus-precision tradeoff — making the threshold explicit highlights this design choice for readers
-
Population structure was visualized using MDS of an IBS distance matrix computed in PLINK↳ Could also: Principal component analysis (PCA) of genotype data is another standard approach for population structure visualization in livestock genomics — PCA has well-characterized statistical properties, is the dominant method in current population genomics practice, and PC loadings relate directly to allele frequencies; MDS and PCA often yield similar two-dimensional layouts, but PCA is more straightforwardly interpretable in a genetic context and integrates naturally with tools such as ADMIXTURE
-
Temporal allele frequency changes were modeled with an HMM approach applied to each line separately against the 1977 population↳ Could also: Likelihood-ratio tests under an explicit Wright–Fisher diffusion model, or Bayesian methods such as BayeScan-Time, could also detect selection from two-time-point allele frequency data — Alternative temporal approaches make different assumptions about the drift and selection coefficient distributions; some additionally provide posterior distributions over selection coefficients, which would complement the point estimates of selective advantage reported here and allow uncertainty quantification
-
The local score approach aggregated SNP-level temporal p-values into candidate regions with a chromosome-wide FPR threshold↳ Could also: Fixed sliding-window aggregation (e.g., mean statistic in 50-kb windows) or permutation-derived window-level thresholds are widely used alternatives in selection-scan studies — Window-based approaches are conceptually simpler and broadly implemented; the local score has the advantage of letting data determine region boundaries without a pre-specified window size, but a fixed-window approach would make the resolution and the number of candidate regions more directly comparable to published livestock selection-scan benchmarks
-
Effective population size was estimated from allele frequency change across two time points using a moment-based estimator (NB package)↳ Could also: Linkage disequilibrium-based Ne estimation or coalescent-based methods (e.g., SMC++) applied within each sample could also be used — LD-based and coalescent methods use within-sample information and can reconstruct Ne trajectories over longer historical periods, providing complementary context; the two-time-point moment estimator is simple and directly interpretable for the sampling interval but does not capture variation in Ne within that period
-
Functional enrichment of candidate-region genes was assessed in Genecodis4 using Homo sapiens as the reference organism to compensate for incomplete pig genome annotation↳ Could also: Tools such as g:Profiler, DAVID, or Enrichr with explicit ortholog mapping to pig or mouse, or permutation-based methods that use the background distribution of gene set sizes in the study organism, could also be applied — Different tools apply different background gene sets, statistical models (hypergeometric, Fisher's exact, or GSEA-style permutation), and annotation database versions; using pig-specific or explicitly mapped annotations would allow direct assessment of whether human-based enrichment results transfer to the study organism's own genomic context
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-36864379
Paper: Boitard et al. 2023, Genet Sel Evol 55:13. "Whole-genome sequencing of cryopreserved resources from French Large White pigs at two distinct sampling times reveals strong signatures of convergent and divergent selection." DOI 10.1186/s12711-023-00789-z · PMCID PMC9979506.
Code (authors' own, P16 N/A): https://github.com/sboitard/LWseq_analysis
(R 61% / shell 26% / python 13%; orchestrated by main.sh).
Data:
- Raw WGS fastq → ENA PRJEB51909 (out of scope: variant calling from raw reads is the wet-lab-adjacent heavy front end, 3 callers × 36 deep WGS genomes).
- Preprocessed PLINK genotypes → Zenodo 10.5281/zenodo.6415023 (the repo's
documented entry point "begin from step 0b"). Two filesets:
snp20_auto_cr.{bed,bim,fam}— HQSNP: autosomal bi-allelic SNPs ≥90% call rate, consensus of samtools+Freebayes+GATK. (.bim 390 MB)all10_auto.{bed,bim,fam}— AV: all variants called by ≥1 method (region characterization input; not needed for the diversity claims)..fam: 36 pigs = 13 female-line (LWD, born 2014–2015) + 13 male-line (LWS, 2012–2016) + 10 ancestral (1977).
Note on scaffold: registry
data_accessionwas harvested asGSE56011. That is the paper's gene-expression GEO series (cited for annotation), NOT the WGS genotypes. The reproducible genotype data live on Zenodo 6415023 + ENA PRJEB51909.
IN SCOPE — deterministic, pipeline-derived from the Zenodo HQSNP genotypes
Run main.sh Section 0/1 pieces on snp20_auto_cr (PLINK 1.9 + base R). All
deterministic (no RNG), minutes of compute.
| id | reported result | paper loc | pipeline |
|---|---|---|---|
| C1 | 13,408,342 autosomal bi-allelic HQSNPs | Results (HQSNP set) | wc -l snp20_auto_cr.bim |
| C2 | 36 animals = 13 LWD + 13 LWS + 10 1977 | Methods / Zenodo | snp20_auto_cr.fam FID counts |
| C3 | polymorphic SNP % per pop: 1977 83.1%, LWD 79.2%, LWS 76.0%; private alleles: 6.9% / 5.3% / 4.3% | Results §"Genetic diversity" | plink --freq --family → scripts/snp_stat.R |
OUT OF SCOPE / not chasing the last 20%
- Ne estimates (LWD≈80, LWS≈74):
estim_Ne.shruns on a random 100k-SNP subset (plink --thin-count 100000, no seed pinned in repo) via a custom temporal-Nb likelihood (estim_NB.R). Stochastic input → not byte-reproducible; best-effort only. - Selection signatures (hapFLK 12 + temporal 12/16 → 38 merged regions, Table 1; 151 genes; convergent/divergent classes): genome-wide hapFLK + compareHMM HMM over 13.4M SNPs — the heavy, multi-tool, longest part. Skipped per 80/20.
- Variant calling from PRJEB51909 raw reads: out of scope (front-end, 3-caller consensus on deep WGS).
Verdict logic
C1/C2 are exact-integer checks. C3 percentages compared within rounding tolerance (paper reports 1 decimal). A clean 1:1 on C1–C3 = faithful reproduction of the diversity layer; selection layer explicitly not attempted.
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.
Clean 1:1 reproduction. All 8 in-scope genetic-diversity claims (HQSNP count 13,408,342; 36-animal composition; polymorphic fractions 83.1/79.2/76.0%; private alleles 6.9/5.3/4.3%) reproduced exactly by running the authors' own PLINK+snp_stat.R pipeline on the authors' own Zenodo genotypes, with deviations only at the rounding level. No fabrication concern — every value is directly derivable from the shipped data. Only caveats are on the data-availability/metadata side (the registry accession GSE56011 was the wrong series, corrected to Zenodo 6415023) and that the paper's heavier selection-signature/Ne layer was explicitly out of scope, so the diversity conclusion is confirmed while the selection conclusion was not tested.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
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.