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

Genetic introgression from commercial European pigs to the indigenous Chinese Lijiang breed and associated changes in phenotypes.

Genet Sel Evol · 2024
64/100 3/4
Why this verdict

The main results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
How its reproducibility compares
64/100
Reproducibility score
0.6 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 25% of all assessed papers rank 854 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 (described well enough; cited tool reproduces exactly). All 3 in-scope pipeline-derived results reproduced on the paper's OWN deposited data (PRJNA942216) via SLURM compute on «our HPC»: C1=VCF2Dis builds + deterministic 203x203 matrix (EXACT, md5 stable across sessions); C2=realized WGS coverage 27.74x autosomal vs reported 27.35x (WITHIN-TOL, +1.4%; DP2a base_count cross-check 28.65x); C3=the full cited pipeline (fastp 1.3.4->BWA 0.7.19->GATK4 4.6.2 with the paper's EXACT hard filters->VCF2Dis->FastME) runs end-to-end on chr18 of 6 real Lijiang samples -> 862,895 PASS SNPs -> real p-distance matrix + NJ tree (STRUCTURAL, not a full-cohort number). OUT OF SCOPE (data-deposition gap, NOT a code failure): every 230-individual headline number (2.41M variants, ADMIXTURE/Fst/D-stats/introgression bins) because 187/230 genomes come from the PigVar DB (unreachable) + IGIGS, none in the deposited accession; and the 44-Lijiang 26.36M-SNP count is the heavy 20% (full joint genotyping) left unrun. Note: a cluster-wide infra event (17:12 UTC) + a flaky period killed two align waves; a «host» supervisor auto-resubmitted (resume from cached fastq) until all 6 aligned cleanly. All compute ran as SLURM jobs on compute nodes per the front1-login-node policy.

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 58
    assessed: 2026-06-19 ⛓ d9ef9ca03b7a
✎ 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
2026-06-22
Rubric version
not recorded
Assessed by
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

Whether commercial European pig (EP) breeds have genetically introgressed into the indigenous Chinese Lijiang pig (LJP) and other Chinese domestic pig breeds, and whether this introgression is associated with phenotypic changes such as thoracic vertebra number.

Core claims
  • Significant genetic introgression from commercial European pigs (EP) into the Lijiang pig (LJP) and other domestic Chinese pig breeds was detected. finding
  • EP introgression is widely prevalent among Chinese domestic pigs but differs significantly between breeds. finding
  • LJP could act as a mediator transmitting EP-introgressed haplotypes to other Chinese pig populations (e.g., Tibetan pigs). mechanism
  • VRTN and STUM are candidate genes underlying the correlation between EP introgression and thoracic vertebra number in LJP. finding
  • A haplotype-based local ancestry approach (Loter) was used to quantify EP introgression proportions across the genome in windows of five SNPs. method
  • EP introgression can be classified into direct (from EP) and indirect (via a mediator population such as LJP) categories. mechanism
  • A combined whole-genome SNP dataset of 230 individuals across Chinese and European pig breeds was generated by merging new LJP resequencing data with public data. resource
Experimental setups
Assay System Perturbation Readout Platform
Whole-genome resequencing 44 unrelated Lijiang pigs (LJP), blood samples none SNP genotypes across the genome Illumina HiSeq X Ten and MGISEQ-2000
D-statistic test, f4 ratio, outgroup-f3 ratio Chinese indigenous pig groups (LJP, SWCDP, TIBP, ECDP, NCDP) vs EP, with Sumatran wild boar as outgroup none Genome-wide admixture/introgression signal (Z-score) Dsuite v0.5 r52; ADMIXTURE v1.3.0
Local ancestry inference (haplotype-based) LJP, NCDP, TIBP-E (target) vs SCDP and EP (reference) none Proportion of genome haplotypes of EP local ancestry per 5-SNP window bin (Z-score >2 = significant) Loter v1.0.1
Phylogenetics and population structure (NJ tree, PCA, DAPC, ADMIXTURE) 230 individuals across 33 pig breeds/groups none Genetic clustering, admixture proportions (K=2-10) PLINK v1.9, VCF2Dis, FastME, GCTA v1.26.0, adegenet, ADMIXTURE v1.3.0
Topology weighting analysis NCDP, TIBP-E, SWCDP, or LJP (P1) vs SCDP (P2) vs EP, Sumatran wild boar as outgroup none Topology weightings per tree window (50 SNPs) Twisst v0.2
Nucleotide diversity (site-pi) analysis SNPs within direct/indirect EP introgression bins and TIBP-derived bins in TIBP-E none Site-pi values compared via Duncan test VCFtools v0.1.16
Phenotype-genotype association (Fisher's exact test, LSBL) 44 LJP split into LJM (>14 thoracic vertebrae, n=27) and LJL (<14 vertebrae, n=17) none (natural phenotypic variation) Differential EP introgression proportion between groups; locus-specific branch length R package fdrtool; custom LSBL calculation
Key results
  • Significant EP introgression detected genome-wide in LJP and other Chinese domestic pig breeds via D-statistic, f4 ratio, and outgroup-f3 analyses Z-score > 3 (D-statistic significance threshold)
  • EP introgression levels vary significantly between different Chinese indigenous pig breeds
  • 51 bins identified as possible indirect EP introgression into TIBP-E via LJP as mediator 51 bins
  • VRTN and STUM identified as candidate genes at loci showing significant introgression differences between high (LJM) and low (LJL) thoracic vertebra number groups FDR < 0.05
  • Final combined dataset comprised 230 individuals with 2,409,528 variants after quality control and imputation 2,409,528 SNPs; marker density 1 SNP/939 bp
  • 44 LJP divided into LJM (n=27, >14 vertebrae) and LJL (n=17, <14 vertebrae) groups for introgression comparison n=27 vs n=17
Key statistics
  • count 27.35x (Average high-quality sequencing depth obtained for LJP samples after fastp filtering)
  • count 38,164,258 SNPs (SNPs remaining after removal of sex-chromosome and unknown-chromosome SNPs, before further QC)
  • count 26,358,572 SNPs (Autosomal SNPs retained after MAF >5%, missing rate <95%, HWE p ≥ 1e-4 filters)
  • count 230 individuals, 2,409,528 variants (Final merged dataset of LJP and public pig genomic data after imputation)
  • count 481,898 bins (Number of 5-SNP window bins used to compute EP introgression proportions genome-wide)
  • other Z-score > 3 (Threshold for significant D-statistic evidence of gene flow between P2 and P3 groups)
  • other Z-score > 2 (Threshold for significant local-ancestry introgression bin (EP or LJP origin))
  • count LJM n=27, LJL n=17 (Grouping of LJP individuals by thoracic vertebra count (>14 vs <14) for introgression-phenotype analysis)

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 whole-genome resequencing study of 44 Lijiang pigs (LJP), integrated with published data from 186 individuals across 33 breeds (230 total), used a suite of population-genetic statistics — D-statistic, f4 ratio, outgroup-f3 ratio, topology weighting (Twisst), and haplotype-based local ancestry inference (Loter) — to characterize and quantify European pig (EP) introgression into Chinese indigenous breeds. Introgression proportions were Z-transformed and compared across populations using kernel density estimation and Kullback–Leibler divergence. The phenotypic association between EP introgression and thoracic vertebra number in LJP was assessed with per-bin Fisher's exact tests (FDR-corrected), genetic diversity differences across introgression categories were evaluated with the Duncan multiple-comparison test, and differentiation was summarized with per-SNP Fst values and locus-specific branch lengths.

Replicationbiological Sample size44 LJP pigs newly sequenced; 186 additional individuals from published datasets; no formal power calculation stated GroupsMultiple Chinese indigenous pig breeds vs. commercial European breeds; within LJP: high-vertebra group LJM (>14, n=27) vs. low-vertebra group LJL (<14, n=17) Pairingunpaired Randomization/blindingnot stated Dispersionnone Effect sizesyes Confidence intervalsno Multiplicity correctionBenjamini-Hochberg FDR via R package 'fdrtool' (for Fisher's exact tests); Duncan test (for site-pi multiple comparisons)
Statistical tests used
Test Applied to n Assumptions
D-statistic (ABBA-BABA) test; significance threshold Z-score > 3 Genome-wide detection of admixture between commercial European pig breeds and each Chinese indigenous pig group, using Sumatran wild boar as outgroup 230 individuals, 2,409,528 phased SNPs not stated
f4 ratio Quantification of EP introgression proportion alongside D-statistic results 230 individuals, 2,409,528 phased SNPs not stated
Outgroup-f3 ratio Detection of EP introgression into each Chinese indigenous pig group relative to the Sumatran wild boar outgroup 230 individuals not stated
Topology weighting (Twisst); 50-SNP sliding windows Topology analysis of EP introgression patterns across NCDP, TIBP-E, SWCDP, and LJP groups relative to SCDP 2,409,528 phased SNPs not stated
Local ancestry inference (Loter) with Z-transformation of introgression proportions; significance threshold Z-score > 2 Haplotype-level quantification of EP introgression in LJP, NCDP, and TIBP-E; also LJP-mediated (indirect) introgression into TIBP-E; 481,898 bins of 5 SNPs 230 individuals across 481,898 5-SNP window bins not stated
Kullback-Leibler (K-L) divergence Comparison of introgression proportion distributions between pig populations (e.g., LJP vs. TIBP-E) within significantly introgressed regions na
Fisher's exact test with FDR correction (R package 'fdrtool'); significance threshold FDR < 0.05 Per-bin comparison of EP introgression proportions between LJM (>14 thoracic vertebrae, n=27) and LJL (<14 thoracic vertebrae, n=17) groups within LJP 44 LJP individuals (LJM n=27, LJL n=17) not stated
Duncan test (multiple comparison) Comparison of average site-pi (nucleotide diversity) across three introgression-category bin sets in TIBP-E: direct EP introgression, indirect EP introgression, and TIBP-derived genomic segments 51 bins per introgression category not stated
Locus-specific branch-length (LSBL) analysis Validation of genetic differentiation in significantly introgressed bins between LJL, LJM, and EP groups 44 LJP individuals; SNPs within detected significant bins not stated
Per-SNP fixation index (Fst) computed by VCFtools Genetic differentiation between TIBP-E and EP, LJP and EP, and TIBP-E and LJP within 51 direct and indirect introgression bins not stated
t-SNE dimensionality reduction with Euclidean distance on t-SNE coordinates Pairwise distances between EP, SWCDP, LJP, NCDP, and TIBP-E based on haplotype frequencies in possible direct and indirect introgression regions 252 haplotypes (indirect) and 356 haplotypes (direct) after frequency filtering not stated
Maximum-likelihood tree (TreeMix) Phylogenetic relationships among pig populations based on SNPs in all possible direct and indirect introgression bins SNPs within 51 bins not stated
Approaches that could also have been used
  • Local ancestry was inferred using Loter with 5-SNP windows and a binary (EP vs. SCDP) reference panel, with hard assignment to one ancestry class
    Could also: RFMix or HAPMIX could also be used for local ancestry inference, providing posterior probability distributions over ancestry states per window rather than hard binary calls — Probabilistic methods provide uncertainty estimates for each window's ancestry assignment, which can be especially informative for recently admixed genomic regions where ancestry is difficult to distinguish; RFMix is also widely benchmarked in livestock admixture mapping contexts
  • Multiple comparisons of average site-pi across three introgression-category groups (direct EP, indirect EP, TIBP-derived) were evaluated with the Duncan test
    Could also: A one-way ANOVA followed by Tukey HSD or a Kruskal-Wallis test followed by Dunn's test with Bonferroni correction could also be used for these three-group comparisons — Tukey HSD controls the family-wise error rate across all pairwise contrasts simultaneously and is more commonly reported in genomics; a nonparametric Dunn's test would be appropriate if site-pi values within bins are not normally distributed, which is plausible for small bin sets
  • The association between EP introgression proportion and thoracic vertebra number was assessed by dividing individuals into two discrete groups (LJM vs. LJL) and applying Fisher's exact test per genomic bin
    Could also: Admixture mapping with logistic or linear regression of vertebra count (or the binary high/low grouping) on per-individual local ancestry proportion at each bin could also be used — Regression-based admixture mapping models the continuous relationship between ancestry proportion and phenotype across individuals rather than testing each bin independently after discretization, and can produce a single genome-wide effect estimate analogous to a GWAS association statistic
  • Kullback-Leibler divergence was used to compare introgression proportion distributions between population pairs within introgressed regions
    Could also: Jensen-Shannon divergence or the Wasserstein (earth-mover's) distance could also be used to compare these distributions — Jensen-Shannon divergence is symmetric and always finite, avoiding KL divergence's asymmetry and undefined values when one distribution assigns zero probability mass; the Wasserstein distance additionally respects the geometry of the proportion scale, which may be informative for continuous introgression values
  • Population structure was characterized using PCA, DAPC, and unsupervised ADMIXTURE (K=2–10 evaluated by cross-validation error)
    Could also: The Evanno delta-K method applied alongside cross-validation error could also provide a complementary heuristic for selecting the optimal K in ADMIXTURE — Cross-validation error and delta-K often suggest different optimal K values; reporting both gives a more complete picture of the range of reasonable population structure models and is a commonly recommended practice in population genomics
  • Genetic differentiation between introgression-category groups was summarized using per-SNP Fst computed by VCFtools
    Could also: Window-based Fst (e.g., sliding 50-kb windows) with smoothing could also be used to summarize differentiation across the 51 introgression bins — Per-SNP Fst is noisy due to stochastic sampling at individual loci; window-averaged Fst reduces this variance and produces a smoother landscape that can be more directly related to the introgression block boundaries already identified by local ancestry inference
Software: fastp 0.23.2 · BWA · GATK 4.0.9 · bcftools 1.8 · Beagle 5.2 · PLINK 1.9 · VCF2Dis 1.50 · FastME 2.1.6.1 · GCTA 1.26.0 · ADMIXTURE 1.3.0 · Dsuite 0.5 r52 · Twisst 0.2 · Loter 1.0.1 · VCFtools 0.1.16 · TreeMix 1.13 · R/ggtree · R/adegenet (DAPC) · R/Pophelper · R/fdrtool

What was reproduced

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

Scope — pmid-38566006

Paper: Yang et al. 2024, Genetic introgression from commercial European pigs to the indigenous Chinese Lijiang breed and associated changes in phenotypes. Genet Sel Evol 56:24. DOI 10.1186/s12711-024-00893-8.

Cited code: https://github.com/hewm2008/VCF2Dis (third-party tool; computes a pairwise p-distance matrix from a VCF, used by the paper to build NJ trees). Deposited data: SRA PRJNA942216.

Critical data-availability finding (governs the whole scope)

The paper analyses 230 individuals. Only the 43 de-novo–sequenced Lijiang pigs (LJL1–17, LJM1–27) are in the deposited accession PRJNA942216 (43 WGS runs, ~1.05 TB raw FASTQ, verified via ENA filereport). The other 187 individuals (European commercial pigs, Tibetan, Eastern/Northern/Southern/SW Chinese pigs, Asian/Sumatran wild boars) come from a separate database ("PigVar") that is not part of the stated accession and for which the paper gives no direct, resolvable download link.

Consequence: the paper's headline numbers are all computed on the merged 230-individual VCF (or subsets of it). That VCF cannot be rebuilt from the deposited accession alone, and full joint genotyping of 230 × 27× WGS pig genomes is far outside an 80/20 reproduction budget. Those results are therefore out of scope (not a code/tool failure — a data-deposition gap).

IN SCOPE (attempted)

DP Result Pipeline / cited tool Feasible because
DP1 The cited code artifact itself VCF2Dis (build + run on its bundled example VCF) → FastME-style NJ output small, self-contained, deterministic; "third-party tool on the paper's data is equally valid" (P16)
DP2 Per-sample WGS coverage vs reported 27.35× average fastp v0.23.2 → BWA-MEM → samtools depth, on a few real PRJNA942216 Lijiang samples uses the paper's own deposited data + the paper's stated aligner
DP3 (stretch) A real NJ distance tree of Lijiang samples via the exact cited pipeline fastp → BWA → GATK4 (paper's exact SNP filters) on a small genomic region → VCF2Dis v1.5x → FastME NJ demonstrates the full cited pipeline runs on the paper's real data; structural (LJL vs LJM), not a full-cohort 1:1 number

OUT OF SCOPE (not attempted — reason recorded)

  • 26,358,572 SNPs / 38,164,258 initial SNPs / 2,409,528 merged variants — require the full 43-sample (resp. 230-sample) joint-genotyped VCF; the 230-set also needs PigVar data not in the accession. Heavy 20%.
  • PCA (GCTA), ADMIXTURE K=2–10, DAPC, Fst, site-pi, TreeMix, Dsuite D/f4/f3, Twisst, Loter local ancestry, introgression-bin counts (14,826 / 72.46 Mb …), vertebra-association haplotype frequencies, LSBL — all computed on the merged 230-individual VCF (PigVar dependency + full-cohort compute). Out of scope.
  • Wet-lab / phenotype measurements (vertebra counts, photographs) — non-pipeline.

Honesty note

No headline numeric claim of the paper is fully reproducible from the deposited accession alone. Expected overall outcome: partial — the cited tool and the front half of the pipeline reproduce cleanly on the paper's real data; the full-cohort population-genetics numbers are blocked by the PigVar data gap, which is itself an auditable finding about the paper's reproducibility surface.

Figures / tables: figures
C1
Reported
VCF2Dis (cited code) builds + computes deterministic pairwise p-distance matrices for the paper's NJ trees
Reproduced
EXACT: built in SLURM «job» (curl main tarball ~v1.56, sh make.sh -O3 -march=native -fopenmp); bundled example Khuman.vcf.gz -> 203x203 p-distance matrix, md5 34a4c36cd7fe743990d1f2deeea82c92, deterministic over 2 runs, matches prior session
exact
C2
Reported
average 27.35x clean coverage (44 de-novo Lijiang pigs)
Reproduced
WITHIN-TOL: DP2b realized BWA->Sscrofa11.1 mapped depth on 6 real PRJNA942216 samples = 27.738x autosomal mean (27.823x whole-genome), mapping ~99.7%; DP2a expected-from-base_count = 28.65x over 44 runs. Delta vs reported = +1.4%
within tolerance
C3
Reported
fastp->BWA->GATK4 hard filters(QD<2,MQ<40,FS>200,SOR>3,MQRankSum<-12.5,ReadPosRankSum<-20)->VCF2Dis->FastME NJ
Reproduced
PARTIAL (structural): full cited pipeline ran end-to-end on real data (chr18, 6 Lijiang samples): 898,848 raw / 862,895 PASS biallelic SNPs -> VCF2Dis 6x6 p-distance matrix (md5 9a032e58b3ed7edc62f459e3fe2e9a8a) -> FastME NJ tree. Not a full-cohort number
partial
C4
Reported
26,358,572 SNPs (38,164,258 initial) in 44 de-novo Lijiang pigs
Reproduced
not attempted (heavy 20%: 44-sample full-WGS joint genotyping ~1.1TB; in principle reproducible from PRJNA942216 alone)
partial
C5
Reported
2,409,528 variants across 230 individuals
Reproduced
not attempted (187/230 genomes from PigVar DB http://«ip»/pigvar/ unreachable + IGIGS; not in deposited accession)
partial
C6
Reported
ADMIXTURE K=6 CV=0.512; D=0.014 Z=2.85; Fst; introgression bins 14826/72.46Mb; TreeMix; Loter; Twisst
Reproduced
not attempted (all on the merged 230-individual VCF; PigVar dependency + full-cohort compute)
partial

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

No assessment has been recorded yet.
🤝
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

<synthetic>

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.

708.5 k
tokens (I/O) · 59.1 M incl. cache
666 min
runtime · 322.82 CPU-h
41.1 GB
peak RAM
9
HPC jobs
hummel
machine