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

Identification of genes influencing the evolution of Escherichia coli ST372 in dogs and humans.

Microb Genom · 2023
L1 89/100 3/4
Why this verdict

The main results reproduced: recomputed values matched the published ones within tolerance.

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 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score 0
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • No relevant deviation in data/preprocessing
  • No authors-side cause for any deviation
  • Any deviation was negligible
  • The central claim held under reproduction
  • Overall, the reproduction was clean
What did not (or only partly)
  • 🟡Reported values were not (fully) derivable from the shared data
How its reproducibility compares
89/100
Reproducibility score
0.8 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 77% of all assessed papers rank 246 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 to reproduce: YES. The repo (CJREID/ST372 @ 4840b5d) ships every downstream intermediate, so the secondary analysis is fully auditable. By an INDEPENDENT Python recompute (not the authors' R script), 7 of 9 in-scope claims reproduce EXACTLY: genome counts 300/72/13/13 (C1), 3,493 core genes (C2), 13 clusters (C4), cluster sizes 202/86/44/22 (C5), 43/15/15 closely-related cross-source SNP pairs (C6), 76 Scoary panGWAS genes split 40/16/11/9 (C7), and fimH9 in 304 genomes/74.7% (C8) -- all 1:1 with the paper. Two are partial: the 22,504 core SNPs and 3.16 Mbp alignment length (C3) are uncheckable because the core-alignment FASTA is not shipped; and running the authors' own monolithic R script end-to-end (C9) is blocked by abricateR version drift, though all the numbers it would produce were reproduced by other means. No mismatches and no fabrication concerns. NOT attempted (out of scope): wet-lab sequencing of the 122 new isolates, full read->assembly of all 407 genomes, IslandViewer web predictions, and the hand-curated M1/M2/M3 genotype/island schematics. One reviewer note: the paper's '76 genes at BH<1E-20' requires the additional non-hypothetical + Odds_ratio>1 (over-represented) filter that the script applies but the text understates.

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.

✎ 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
2026-06-30
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-30
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

It is unknown whether Escherichia coli ST372 strains from dogs and humans represent shared or distinct populations, and what genomic traits might explain the prominence of ST372 in dogs or its presence in humans.

Core claims
  • Dogs are the dominant host of E. coli ST372, and clusters within the ST372 population structure exhibit distinctive O:H types. finding
  • Cluster M, comprising almost half of the collection, contains two divergent human-restricted clades carrying different O:H types than the rest of the cluster. finding
  • There is evidence of transmission of ST372 between dogs and humans within different phylogenetic clusters, including cluster M. finding
  • Multiple independent acquisitions of the pdu propanediol utilization operon occurred in clusters dominated by canine-source isolates, possibly linked to diet. finding
  • Loss of the pdu operon and acquisition of K antigen virulence genes characterize human-restricted ST372 lineages. finding
  • O:H type generally corresponds tightly with phylogenetic cluster but not with host source, while fimH allele shows no correspondence with cluster. finding
  • A global collection of 407 E. coli ST372 whole-genome sequences was assembled and analysed to define population structure, O:H types and accessory genome associations. resource
  • Pan-genome-wide association study (Scoary) with a Benjamini-Hochberg adjusted P-value cutoff of 1E-20 was used to identify genes associated with fastbaps clusters. method
Experimental setups
Assay System Perturbation Readout Platform
Whole-genome sequencing (Illumina, Hackflex library prep) E. coli ST372 isolates from dogs, humans, wildlife and environment (n=122 newly sequenced) none genome sequence/assembly Illumina NovaSeq S4 (Novogene); MiSeq V2 Nano QC
Core genome maximum-likelihood phylogenetics 407 ST372 genomes plus ST127 outgroup strain SRR5336297 none phylogenetic tree topology and clustering IQ-TREE 2.0.3, GTR+F+R model, 1000 bootstrap replicates
Population structure clustering (fastbaps) core gene alignment of 407 ST372 genomes none assignment of isolates to 13 clusters (A-M) Fastbaps with 'baps' prior
Pairwise core SNP distance analysis core SNP alignment (22,504 bp) from 407 ST372 genomes none pairwise SNP distances between isolates snp-sites 2.5.1, snp-dists 0.6.3
Pan-genome-wide association study (panGWAS) ST372 pan-genome vs fastbaps cluster membership none genes significantly associated with clusters (adjusted P-value) Scoary 1.6.16
In silico gene screening draft genome assemblies of 407 ST372 isolates none presence/absence of AMR, virulence, plasmid replicon, serotype and IS genes ABRicate 1.0.1 against CARD, VFDB, PlasmidFinder, SerotypeFinder, ISFinder, custom ColV/virulence DB
Genomic island prediction selected ST372 genomes carrying GWAS-identified genotypes vs ST127 reference ECONIH2 none predicted genomic island boundaries and content IslandViewer 4
In silico MLST, serotyping and fimH typing draft genome assemblies of 407 ST372 isolates none ST confirmation, O:H type, fimH allele, pMLST plasmid type MLST 2.19.0, SerotypeFinder, pMLST (CGE tools), PointFinder
Key results
  • Cluster M was the largest phylogenetic cluster, comprising 202/407 (49.6%) of sequences 49.6%
  • Cluster M contained most canine sequences (120/202) and most human sequences in the collection (56/72 human total, 77.8%) 56/72 (77.8%)
  • Canine-source sequences dominated the overall study collection 300/407 (73.7%)
  • 29 human sequences and 1 environmental sequence formed a divergent human-dominated clade within cluster M carrying O18:H31, distinct from the O83:H31-dominant remainder of the cluster 29 sequences
  • Clusters G, L and J were canine-dominated but also contained human sequences, each associated with a distinct O:H type (O4:H31, O15:H31, O117:H28 respectively) shared between canine and human isolates
  • fimH9 was the major fimH allele identified across all sources and clusters 304/407 (74.7%)
  • Multiple acquisitions of the pdu propanediol utilization operon occurred in canine-dominated clusters
  • Human-restricted lineages showed loss of the pdu operon and acquisition of K antigen virulence genes
Key statistics
  • count 407 ST372 whole-genome sequences (285 public, 122 newly generated) (total study collection size)
  • count 300 canine-source sequences (73.7%) (source distribution of collection)
  • count 72 human-source sequences (17.7%) (source distribution of collection)
  • count 202/407 (49.6%) (size of cluster M)
  • count 56/72 human sequences (77.8%) in cluster M (human sequence enrichment in cluster M)
  • pvalue Benjamini-Hochberg adjusted P-value cutoff of 1E-20 (significance threshold for Scoary gene-cluster associations)
  • count 304/407 (74.7%) carried fimH9 allele (dominant fimH allele frequency)
  • other 3,160,664 bp core gene alignment (3493 genes); 22,504 bp core SNP alignment (basis for phylogenetic and SNP distance analyses)

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 genomic epidemiology study of 407 E. coli ST372 whole-genome sequences that relies primarily on bioinformatic and phylogenetic methods rather than classical inferential statistics: population structure was defined via a maximum-likelihood core-genome phylogeny (IQTree, bootstrap support) and Bayesian clustering (fastbaps), and a pan-genome-wide association study (Scoary) with a Benjamini-Hochberg-adjusted p-value cutoff was used to link genes to cluster membership. Most other results (host/source, O:H type, fimH allele distributions, SNP distances) are reported descriptively as counts and percentages or pairwise distance summaries rather than through explicit hypothesis tests.

Replicationunclear Sample sizeTotal collection of 407 genomes (285 public, 122 newly generated) described by source, continent, and cluster with counts and percentages (e.g., cluster M n=202, 49.6%); no power or sample-size calculation stated Groupsphylogenetic clusters (A-M) and sources/hosts (canine, human, wildlife, environmental) compared by O:H type, fimH allele, and accessory gene content Pairingna Randomization/blindingna Dispersionnone Multiplicity correctionBenjamini-Hochberg FDR adjustment (implemented within Scoary)
Statistical tests used
Test Applied to n Assumptions
Scoary pan-genome-wide association (gene-trait association test) association between fastbaps cluster membership and pan-genome gene presence/absence not stated
Maximum-likelihood phylogenetic inference with non-parametric bootstrap support (IQTree, GTR+F+R model, 1000 replicates) core-genome phylogeny of ST372 (Fig. 2) 3,160,664 bp core gene alignment of 3493 genes from all ST372 sequences plus an ST127 outgroup not stated
Fastbaps Bayesian clustering (BAPS prior) definition of population structure clusters A-M core gene alignment and maximum-likelihood tree from all ST372 sequences (n=407) not stated
Approaches that could also have been used
  • Gene-cluster associations were identified using Scoary with a Benjamini-Hochberg-adjusted p-value cutoff of 1E-20.
    Could also: Bacterial GWAS tools that explicitly model or correct for population structure/clonal relatedness, such as pyseer with a linear mixed model or phylogeny-aware fixed-effects correction — this can help distinguish gene-trait associations arising from a direct functional link versus those arising simply through shared ancestry (lineage effects), a common consideration in bacterial GWAS.
  • Population structure was defined using fastbaps clustering applied to the core-genome alignment and tree.
    Could also: Complementary clustering approaches such as hierBAPS, PopPUNK, or recombination-aware clustering (e.g., ClonalFrameML) — cross-checking cluster assignments with an independent method, or one that explicitly accounts for recombination, can provide additional support for the boundaries drawn between lineages.
  • Phylogenetic relationships were inferred from a single maximum-likelihood tree (IQ-TREE, GTR+F+R model) with 1000 bootstrap replicates for node support.
    Could also: A recombination-aware tree-building step (e.g., Gubbins or ClonalFrameML) applied to the core alignment prior to or alongside the ML tree, or a Bayesian phylogenetic approach (e.g., BEAST) — recombination is common in bacterial core genomes and accounting for it can refine branch-length estimates and topology, complementing bootstrap-based confidence in the ML tree.
  • The Benjamini-Hochberg-adjusted p-value cutoff for Scoary associations was set at a very low value (1E-20), described partly as a way to manage the number of candidate genes.
    Could also: Reporting effect sizes (e.g., odds ratios) and confidence intervals for top gene-cluster associations alongside the adjusted p-value threshold — pairing a significance threshold with an effect-size estimate lets readers gauge both the statistical evidence and the practical strength of each association.
  • Distributions of source/host, O:H type, and fimH allele across phylogenetic clusters were summarized descriptively as counts and percentages (Fig. 1).
    Could also: Formal categorical association tests such as chi-squared or Fisher's exact tests (with a multiplicity correction across the multiple category comparisons) — a formal test statistic and p-value would complement the descriptive percentages, offering a quantitative measure of how strongly source, O:H type, or fimH allele associate with cluster.
  • Relatedness between canine and non-canine sequences was assessed via pairwise core-genome SNP distances (matrix from snp-dists).
    Could also: Summarizing within- versus between-cluster SNP-distance distributions (e.g., median and IQR) or applying a defined SNP-distance threshold with a transmission-cluster inference method — this could offer a more formal, quantitative summary of relatedness signal to complement the pairwise distance matrix.
Software: Scoary 1.6.16 · IQ-TREE 2.0.3 · Fastbaps · Roary 3.13.0 · snp-dists 0.6.3 · R/RStudio (custom script) R 4.1.3 / RStudio 1.4.1106

What was reproduced

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

Scope — pmid-36752777

Paper: Elankumuran et al. 2023, Microb Genom 9(2):000930. "Identification of genes influencing the evolution of Escherichia coli ST372 in dogs and humans." Code: https://github.com/CJREID/ST372 (authors' own; branch main). Data: SRA BioProjects PRJNA678027 (MVC genomes) + PRJNA827950 (collaborator); 407 ST372 genomes total (285 public + 122 newly sequenced).

What the repo ships (key for scope)

The repo ships all downstream intermediate files under data/, so the entire secondary analysis is reproducible without re-running the upstream read→assembly steps:

  • data/roary/gene_presence_absence.Rtab (15.7 MB) — Roary pangenome matrix (0/1)
  • data/roary/ST372.roary.meta.csv — pangenome + metadata join
  • data/iqtree/core_gene_alignment.OGremoved.contree — ML phylogeny (consensus tree)
  • data/fastbaps/ST372.fastbaps.csv — fastbaps cluster assignments
  • data/snps/CGA.snp-dists.csv — pairwise core-genome SNP distance matrix
  • data/scoary/cluster/ — 13 Scoary panGWAS result files (one per cluster)
  • data/meta/ST372_Metadata_*.csv, MVC_specimens.csv — sample metadata
  • data/summaries/ — mlst.txt, ST372_serotype.tsv, ST372.fimH.csv, genotype.txt, assembly_stats.txt, bracken_report.txt, pMLST.txt, pointfinder.txt, ST372.pUTI89.tab
  • scripts/ST372_analysis.R (53 KB) — produces all figures + supplementary tables.

IN SCOPE (pipeline-derived; we attempt)

# Reported result Pipeline / source file Strategy
C1 407 genomes total (300 canine, 72 human, 13 wild, 13 env) metadata + Rtab columns count
C2 Core genome = 3,493 genes; 3,160,664 bp alignment Roary gene_presence_absence.Rtab recompute core-gene count at Roary thresholds
C3 22,504 core SNP sites snp-sites on core alignment (alignment NOT shipped) partial — cross-check via snp-dists
C4 13 fastbaps clusters (A–M) ST372.fastbaps.csv count unique clusters
C5 Cluster sizes: M=202 (49.6%), G=86 (21.1%), L=44 (10.8%), J=22 (5.41%) fastbaps + metadata count per cluster
C6 43 closely-related canine–human pairs ≤30 SNPs CGA.snp-dists.csv + source metadata recompute pairwise
C7 76 Scoary-significant genes (BH-adjusted P<1E-20); per-cluster 40/16/11/9 (M/G/L/J) data/scoary/cluster/* recompute threshold count
C8 Serotype O83:H31 dominant in M (167/202); fimH9 most prevalent (304, 74.7%) serotype.tsv, fimH.csv recompute
C9 Authors' ST372_analysis.R runs and regenerates figures/tables R 4.1.x + packages execute end-to-end

OUT OF SCOPE (not attempted; reasons)

  • Wet-lab: DNA extraction, Illumina sequencing of the 122 new isolates (lab work).
  • Full upstream read→assembly (fastp→Kraken2→Shovill) for 407 genomes: raw assemblies/ the core alignment FASTA are not shipped; re-deriving the 3.16 Mbp alignment + 22,504 SNPs would require assembling all 407 (massive, and depends on exact public-genome set). We treat C3 as partial (cross-checked, not re-derived). A subset assembly of the newly-sequenced PRJNA678027 reads is feasible as a dataset-profiling spot check.
  • IslandViewer 4 predictions (web service, not scriptable deterministically).
  • Manual genotype/island schematic curation (M1/M2/M3 hand-defined genotypes).
Figures / tables: Fig 1Table
C1
Reported
407 genomes: 300 canine, 72 human, 13 wild, 13 environmental
Reproduced
Canine 300, Human 72, Wild 13, Env 13 (+Livestock 6, Aquatic 3); total 407
exact
C2
Reported
3,493 core genes (3,160,664 bp alignment)
Reproduced
3,493 genes present in >=99% of 408 genomes (>=404)
exact
C3
Reported
22,504 core SNP sites
Reproduced
uncheckable from shipped data (core alignment FASTA not in repo)
partial
C4
Reported
13 fastbaps clusters (A-M)
Reproduced
13 Level-1 clusters
exact
C5
Reported
M=202, G=86, L=44, J=22
Reproduced
202, 86, 44, 22
exact
C6
Reported
43 canine-human, 15 canine-wild, 15 canine-environment pairs <=30 core SNP
Reproduced
43, 15, 15
exact
C7
Reported
76 Scoary genes BH<1E-20 (M40/G16/L11/J9)
Reproduced
76 (M40/G16/L11/J9) with BH<1e-20 + non-hypothetical + Odds_ratio>1
exact
C8
Reported
fimH9 most prevalent: 304 (74.7%)
Reproduced
304/407 (74.7%)
exact
C9
Reported
authors' ST372_analysis.R regenerates figures/tables
Reproduced
env+packages build; script halts mid-run (abricateR ColV 'Sequence_type' version drift)
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 89/100

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 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score 0

This is a strong, near-1:1 reproduction: 7 of 9 in-scope claims reproduce exactly by an independent Python recompute from the authors' deposited intermediate files (CJREID/ST372 @ 4840b5d), with no mismatches and no fabrication signal. The two partials are both benign — C3's 22,504 SNPs and 3.16 Mbp alignment length are uncheckable only because the core-alignment FASTA was not shipped (data-availability, not authors' defect), and C9's failure to run the monolithic R script end-to-end stems from abricateR version drift. The one reporting nit is that the '76 genes at BH<1E-20' headline (C7) actually needs the full non-hypothetical + Odds_ratio>1 filter to land exactly, an understated-method issue worth noting but not a discrepancy. Overall quality is green with q5 marked yellow purely because two values are not derivable from the shared data.

🤝
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.