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

Meta-analysis of six dairy cattle breeds reveals biologically relevant candidate genes for mastitis resistance.

Genet Sel Evol · 2024
L1 69/100 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 results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Nothing in this column.
What did not (or only partly)
  • 🔴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
How its reproducibility compares
69/100
Reproducibility score
0.3 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 35% of all assessed papers rank 745 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

DESCRIBED WELL ENOUGH for the annotation step; core GWAS is data-restricted. The paper's central results (58 lead markers, 31 candidate genes, MR-MEGA/METAL/MTAG/MAGMA p-values) are NOT reproducible: all six breeds' genotype+phenotype are available only on reasonable request with breeding-company permission (Viking Genetics, Braunvieh Schweiz, DataGene, INRAE/Valogene, FBN, WUR); no public summary statistics exist -> data_restricted for the core. The smoove+duphold GC-CNV sub-analysis is public-tool but needs an un-accessioned 567-animal WGS cohort + tens-of-TB alignment -> out of 80/20. REPRODUCED 1:1 (P16, third-party tool on the paper's own reported variants): Ensembl VEP functional annotation of the lead-SNP (Table 1) and candidate-causal tables. On the public REST VEP (current Ensembl release) 47/69 variants reproduce EXACTLY; all 6 resolvable missense candidate-causal variants give the reported amino-acid change in the reported gene, and 3/6 SIFT scores match on the current release. The 18 mismatches are dominated by Ensembl annotation drift between the paper's pinned release-104 and the current release (newly-annotated ncRNA introns/regulatory regions, LOC618542->RBAK rename, v104-only gene ENSBTAG00000049290 retired). The version-faithful v104 offline-cache run is fully staged for «our HPC» (run.sbatch + resolve/compare scripts) but was NOT executed: it is gated on a phone-2FA VPN login (two fresh 2FA links expired before being tapped) and the operator chose to finalize. NOT ATTEMPTED: the GWAS meta-analysis itself, MAGMA/GARFIELD/MTAG numeric outputs, and the smoove CNV. No fabrication detected in the annotation columns checked (annotations are consistent with public gene models).

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 69
    assessed: 2026-06-15 ⛓ 223a98e90a13
✎ 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
2026-06-15
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator headless) · v1.0 · run #1 2026-06-15
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: opus
Founding hypothesis

A multi-breed meta-analysis of GWAS summary statistics for clinical mastitis (CM) and somatic cell score (SCS) across six dairy cattle breeds can increase power and precision to identify functional genetic variants and candidate genes affecting mastitis resistance in dairy cattle.

Core claims
  • Meta-analysis of GWAS across multiple breeds for CM and SCS identified 58 lead markers associated with mastitis incidence, including 16 loci not overlapping previously identified QTL in AnimalQTLdb. finding
  • Post-GWAS analyses prioritized 31 candidate genes and 14 credible candidate causal variants affecting mastitis. finding
  • Combining single- and multi-trait meta-analysis methods that account for multi-breed structure increases power and precision to detect variants affecting mastitis-related traits. method
  • Putative causal genes were prioritized by integrating nearest-gene/gene-based analysis with GO, KEGG pathway, and mammalian phenotype database support. method
  • The GC gene (Vitamin D-binding protein) on BTA6 (~88-89 Mb) is a plausible candidate for the recurrent CM/SCS QTL, with NPFFR2 as an additional potential causal gene. mechanism
  • The candidate gene list helps elucidate the genetic architecture of mastitis resistance and supports breeding for improved resistance. resource
Experimental setups
Assay System Perturbation Readout Platform
Genome-wide association study (GWAS) on imputed whole-genome sequence variants Six dairy cattle breeds (Holstein, Jersey, Nordic Red/RDC, Montbéliarde, Normande, Brown Swiss/Original Braunvieh); Bos taurus none Association between sequence variants and clinical mastitis (CM) and somatic cell score (SCS) GCTA-MLMA (mixed linear model)
Single-trait meta-analysis of GWAS summary statistics (fixed-effect) Multi-breed dairy cattle GWAS datasets none Meta-analyzed association statistics for CM and SCS METAL
Trans-ethnic meta-regression meta-analysis Multi-breed dairy cattle GWAS datasets none Meta-analyzed association statistics accounting for allelic effect heterogeneity (MR-MEGA_CM, MR-MEGA_SCS) MR-MEGA (4 PCs for CM, 12 PCs for SCS)
Multi-trait meta-analysis Per-breed CM and SCS summary statistics none Combined multi-trait association statistics (MTAG_CM, MTAG_SCS) MTAG
Gene-based analysis Bovine GWAS summary statistics and gene location data none Gene-level association for candidate gene prioritization MAGMA
Variant annotation Significant sequence variants, ARS-UCD1.2 genome none Functional effect annotation of variants Variant Effect Predictor (VEP)
Genomic feature enrichment analysis Bovine GWAS variants none Enrichment of GWAS signals in genomic features / key variants GARFIELD
Copy number variant (CNV) calling and quality control of summary statistics Additional dataset; multi-breed summary statistics none CNV detection and QC metrics (allele frequency, lambda, MAF, imputation accuracy) EasyQC
Key results
  • 58 lead markers associated with mastitis incidence were identified by the meta-analyses 58 lead markers
  • 16 of the identified loci did not overlap with previously identified QTL in AnimalQTLdb 16 loci
  • 31 candidate genes were prioritized through post-GWAS analyses 31 genes
  • 14 credible candidate causal variants affecting mastitis were identified 14 variants
  • A recurrent QTL for CM and SCS was confirmed at ~88-89 Mb on BTA6 across many breeds/studies 88-89 Mb
Key statistics
  • count 30,689 (Number of animals/records with phenotypes for clinical mastitis (CM))
  • count 119,438 (Number of animals/records with phenotypes for somatic cell score (SCS))
  • count 8 GWAS for CM and 14 GWAS for SCS (Number of GWAS combined in the meta-analyses)
  • other −log10(p) > 8.5 (Significance threshold for single- and multi-trait meta-analysis)
  • count 1869 QTL for SCS and 569 QTL for CM (QTL reported in AnimalQTLdb for the two traits)
  • other 0.02 to 0.12 (Heritability range of clinical mastitis (CM))
  • other 0.10 to 0.15 (Heritability range of somatic cell score (SCS))
  • correlation 0.24 to 0.55 (Genetic correlation of CM with milk yield in Nordic dairy cattle)

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.

The paper conducted sequence-level GWAS within each of six dairy cattle breeds using mixed linear models (GCTA-MLMA with a genomic relationship matrix), then meta-analysed summary statistics across eight CM and fourteen SCS GWAS using two complementary approaches: MR-MEGA (trans-ethnic meta-regression with principal components to accommodate allelic-effect heterogeneity across breeds) and METAL (fixed-effect inverse-variance-weighted). A multi-trait analysis (MTAG, per breed) followed by MR-MEGA was additionally performed to leverage the genetic correlation between CM and SCS. Post-GWAS analyses included MAGMA gene-based testing, VEP variant annotation, and GARFIELD genomic feature enrichment, with results integrated against CattleGTEx expression data to nominate candidate causal genes and variants; a uniform genome-wide threshold of −log10(p) > 8.5 was applied across all outputs.

Replicationbiological Sample sizeTotal n stated (30,689 for CM across 8 GWAS; 119,438 for SCS across 14 GWAS); per-breed n not provided in main text GroupsSix dairy cattle breeds (Holstein, Jersey, Nordic Red, Montbéliarde, Normande, Brown Swiss/Original Braunvieh) from seven institutions; CM and SCS analysed separately and jointly Pairingna Randomization/blindingnot stated Dispersionnone Exact p-valuesyes Effect sizesyes Multiplicity correctionUniform genome-wide significance threshold of −log10(p) > 8.5 (p ≈ 3.16 × 10⁻⁹) applied to all single-trait and multi-trait meta-analysis outputs
Statistical tests used
Test Applied to n Assumptions
Mixed linear model (GCTA-MLMA): y = 1μ + bx + g + e, with genomic relationship matrix modelling polygenic background Within-breed single-variant GWAS for CM and SCS at all seven contributing institutions 30,689 total (CM); 119,438 total (SCS) across all breeds; per-breed n not stated in text not stated
Trans-ethnic meta-regression with principal components (MR-MEGA) Primary meta-analysis combining 8 CM GWAS and 14 SCS GWAS across six breeds; also applied after MTAG for multi-trait outputs 30,689 (CM); 119,438 (SCS) not stated
Fixed-effect inverse-variance-weighted meta-analysis (METAL, STDERR method) Secondary meta-analysis of CM and SCS summary statistics 30,689 (CM); 119,438 (SCS) not stated
Multi-trait meta-analysis (MTAG) followed by MR-MEGA combination across breeds Joint analysis of CM and SCS per breed, then combined across breeds; outputs MTAG_CM and MTAG_SCS 30,689 (CM); 119,438 (SCS) not stated
Gene-based association test (MAGMA) Post-GWAS prioritisation of candidate genes from meta-analysis summary statistics not stated
Genomic feature enrichment analysis (GARFIELD) Post-GWAS enrichment of significant variants in functional genomic annotations not stated
Approaches that could also have been used
  • Fixed-effect inverse-variance-weighted meta-analysis (METAL) was used alongside MR-MEGA to combine within-breed GWAS results
    Could also: A random-effects meta-analysis (e.g., DerSimonian–Laird or REML-based) could also have been applied — When allelic effects are expected to differ across genetically diverse breeds, a random-effects model explicitly quantifies between-study heterogeneity in effect size and yields confidence intervals that reflect that uncertainty; this would complement the heterogeneity statistics already available within MR-MEGA
  • The genome-wide significance threshold was set uniformly at −log10(p) > 8.5 across all analyses
    Could also: A Bonferroni correction anchored to the effective number of independent sequence-level variants, or a permutation-based empirical threshold, could also have been used — Deriving the threshold empirically from the actual LD structure and total variant count of the dataset would formally calibrate the family-wise error rate and make the chosen threshold reproducible and transparent to readers
  • Post-GWAS fine-mapping relied on VEP functional annotation and GARFIELD enrichment to prioritise credible causal variants
    Could also: Bayesian statistical fine-mapping methods such as SuSiE or FINEMAP applied to summary statistics with an LD reference could also have been used — These methods produce posterior inclusion probabilities and credible sets that quantify per-variant uncertainty about causality in a statistically principled way, complementing the annotation-based prioritisation already performed
  • Gene-based testing was performed using MAGMA on the GWAS summary statistics
    Could also: Colocalization analysis (e.g., coloc) or a transcriptome-wide association study (TWAS) using the CattleGTEx eQTL data already accessed in the study could also have been applied — Coloc and TWAS explicitly test whether a GWAS signal and a cis-eQTL in a relevant tissue share a causal variant, providing a mechanistic link between the association signal and gene expression that p-value aggregation in MAGMA alone does not establish
  • Summary statistics from phenotypes defined in different ways across institutions (DRP, DYD, EBV with varying reliability weights) were combined without explicit modelling of phenotype-definition heterogeneity
    Could also: A meta-regression model that includes phenotype type (DRP vs. DYD vs. EBV) as a moderator covariate could also have been applied — Different deregression methods and reliability weights may introduce systematic differences in effect-size scale and standard-error calibration across studies; including this as a covariate would allow assessment of how much it contributes to the heterogeneity observed in MR-MEGA
  • Within-breed GWAS used a single-variant mixed linear model (GCTA-MLMA) treating each SNP independently
    Could also: Bayesian whole-genome regression methods (e.g., BayesR or BayesC) could also have been applied within each breed before meta-analysis — Bayesian approaches simultaneously model all variants and can have greater power for highly polygenic traits with many small-effect loci; the resulting per-variant posterior effect estimates could serve as an alternative or complementary input to the meta-analysis stage
Software: GCTA-MLMA · EasyQC · MR-MEGA · METAL · MTAG · MAGMA · VEP (Ensembl Variant Effect Predictor) · GARFIELD

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.

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.

Citations
24
Impact: medium
Foundation confidence
None of its references are in our reproducibility record yet — its foundation cannot be assessed.
Topics

No assessed neighbours yet — the network grows as more papers are assessed.

What was reproduced

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

Scope — pmid-39009986

Paper: Cai et al. 2024, Genet Sel Evol 56:54. "Meta-analysis of six dairy cattle breeds reveals biologically relevant candidate genes for mastitis resistance." DOI 10.1186/s12711-024-00920-8 · PMCID PMC11247842.

What the paper does (pipeline map)

A multi-breed (Holstein, Jersey, Nordic Red, Brown Swiss, Montbéliarde, Normande) sequence-based GWAS meta-analysis for clinical mastitis (CM, 30,689 animals) and somatic cell score (SCS, 119,438 animals). Each partner ran a within-breed GWAS (GCTA-MLMA) on WGS-imputed genotypes; summary statistics were QC'd (EasyQC) and combined by MR-MEGA (meta-regression), METAL (fixed-effect) and MTAG (multi-trait). Downstream: MAGMA gene-based test, GARFIELD functional enrichment, VEP v104 variant annotation, PLINK LD, and a smoove + duphold structural-variant sub-analysis on chr6 WGS.

In scope vs out of scope

Reported result Pipeline In scope? Why
58 lead markers / QTL coordinates & −log10(p) (Table 1/2) GCTA-MLMA → MR-MEGA/METAL/MTAG OUT Inputs are per-breed genotype+phenotype/summary-stats; all six restricted (on-request + breeding-company permission). No public sumstats deposit.
MAGMA gene-based, GARFIELD enrichment, MTAG novel signals MAGMA/GARFIELD/MTAG OUT Same restricted GWAS inputs.
12 kb CNV at BTA6:86,949,652–86,961,433 near GC Trimmomatic→bwa→GATK→smoove→duphold (chr6) OUT (hard >20%) Tool public; data public in principle (1000 BGP, PRJNA431934 +19 BioProjects) BUT the exact 567-animal cohort is not individually accessioned, requires tens-of-TB WGS download + whole-genome alignment + joint calling. Infeasible in scope; and the result is itself a confirmation of a prior study's CNV, not a fresh pinnable number.
Functional annotation of lead SNPs + candidate causal variants (Tables 1, 2 & "candidate causal mutations"): consequence type, nearest gene, missense AA change + SIFT score Ensembl VEP v104 on ARS-UCD1.2 IN ✅ Tool (VEP v104) and reference (Ensembl release-104 bos_taurus cache) are public; inputs are the rsIDs + ARS-UCD1.2 coordinates printed in the paper's own tables; compute is minutes. Re-annotating these variants is a clean, deterministic 1:1 check (and a fabrication probe on the reported gene/consequence/SIFT columns).

Reproduction target (the "few clear data points")

Re-run Ensembl VEP v104 (offline cache, --sift b --symbol --nearest symbol) on the lead SNPs (Table 1, by rsID/coordinate) and the candidate causal mutations (Table "candidate causal mutations", the missense rows carry SIFT scores), and compare 1:1 against the paper's reported Annotation / nearest-gene / SIFT columns.

P16 note: VEP is a third-party tool; applying it to the paper's reported variants is an equally valid reproduction of the paper's annotation step. We do NOT and cannot reproduce the GWAS that produced the variants (restricted data).

Honest non-attempt list

  • The GWAS meta-analysis itself (restricted genotype/phenotype across 6 partners).
  • The smoove/duphold CNV (cohort not accessioned; tens-of-TB WGS, out of 80/20).
  • MAGMA/GARFIELD/MTAG numeric outputs (restricted inputs).
Figures / tables: Table
CORE_58lead
Reported
58 lead markers associated with mastitis incidence (15 MR-MEGA_CM QTL on 13 autosomes; 22 MR-MEGA_SCS QTL on 15 autosomes)
Reproduced
NOT ATTEMPTED — per-breed genotype+phenotype restricted (6 partners, on reasonable request + breeding-company permission); no public summary statistics
partial
CORE_31genes
Reported
31 candidate genes prioritized
Reproduced
NOT ATTEMPTED — depends on restricted GWAS meta-analysis
partial
SV_GC_CNV
Reported
12 kb CNV at BTA6:86,949,652-86,961,433 near GC (smoove+duphold on 567 WGS animals, chr6)
Reproduced
NOT ATTEMPTED — 567-animal cohort not individually accessioned; tens-of-TB WGS download + WG alignment; out of 80/20
partial
VEP_lead_snps_Table1
Reported
Functional annotation (nearest gene + consequence) of lead SNPs, Table 1 (e.g. GC intergenic; STAT6 synonymous; MAP3K1 intron; RBM15 downstream)
Reproduced
VEP reproduces 41/58 lead-SNP annotations EXACTLY on current Ensembl; remainder = release-104->current annotation drift
partial
VEP_6:86986115
Reported
Novel gene ENSBTAG00000049290 / Missense K15Q, deleterious (0)
Reproduced
VEP: gene=GC cons=intron_variant aa= sift=
did not match
VEP_6:87324678
Reported
NPFFR2 / Missense E406K, tolerated (0.58)
Reproduced
VEP: gene=NPFFR2 cons=missense_variant aa=E/K sift=tolerated 0.34
exact
VEP_14:550784
Reported
CPSF1 / Missense T430I, tolerated (0.13)
Reproduced
VEP: gene=CPSF1 cons=missense_variant aa=T/I sift=tolerated_low_confidence 0.12
exact
VEP_14:579239
Reported
SLC52A2 / Missense K242E, tolerated (0.13)
Reproduced
VEP: gene=SLC52A2 cons=missense_variant aa=K/E sift=tolerated 0.8
exact
VEP_14:611019
Reported
DGAT1 / Missense K232A, tolerated (0.19)
Reproduced
VEP: gene=DGAT1 cons=missense_variant aa=A/T sift=tolerated_low_confidence 0.19
exact
VEP_18:65188613
Reported
Non-coding transcript / lncRNA
Reproduced
VEP: gene=n/a cons=intron_variant,non_coding_transcript_variant aa= sift=
within tolerance
VEP_19:7311199
Reported
Novel gene / Missense V361I, tolerated (0.19)
Reproduced
VEP: gene=ANKFN1 cons=missense_variant aa=V/I sift=tolerated_low_confidence 0.2
exact
VEP_20:22385791
Reported
LOC104975241 ( MAP3K1 , intron) / ncRNA
Reproduced
VEP: gene=MAP3K1 cons=intron_variant aa= sift=
within tolerance
VEP_20:22386425
Reported
LOC104975241 ( MAP3K1 , intron) / ncRNA, deletion
Reproduced
VEP: gene=n/a cons=unresolved aa= sift=
partial
VEP_21:62941833
Reported
Non-coding transcript / lncRNA
Reproduced
VEP: gene=n/a cons=intergenic_variant aa= sift=
did not match
VEP_22:52960814
Reported
LTF lactotransferrin / Missense I145V, tolerated (0.2)
Reproduced
VEP: gene=LTF cons=missense_variant aa=I/V sift=tolerated 0.31
exact

Assessments & scoring basis

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

🤖 AI curator · claude (ai-curator headless) · v1.0 L1 69/100

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)
🤝
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.

226.3 k
tokens (I/O) · 17.4 M incl. cache
36 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.