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

Chemical and genomic characterization of a potential probiotic treatment for stony coral tissue loss disease.

Commun Biol · 2023
L1 85/100 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: 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 +2
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • Reported values are derivable from the shared data
  • Any deviation was negligible
  • The central claim held under reproduction
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
  • 🟡A deviation was attributed to the published material
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
85/100
Reproducibility score
0.6 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 67% of all assessed papers rank 348 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

STRONG PARTIAL reproduction with real compute on «our HPC». 16S amplicon pipeline reproduces 1:1: deterministic downstream EXACT (ps1/2/5/10=148/72/21/9 ASVs, 104/57/16/6 genera, 263 genera); full DADA2 from shipped reads gives byte-identical read-tracking and 447->429 ASVs EXACT. KEY AUDIT FINDING: the paper's ASV/genera counts reproduce ONLY when the authors' DADA2 merge BUG is replicated (run NS2051 read twice, NS1995 dropped); a corrected merge yields 1003->962 ASVs / 484 genera. So the published numbers are GENUINE outputs of the shipped (defective) code, NOT fabricated. Genome (SPAdes 3.13.0): total length 5,134,546 bp vs 5,134,549 (3 bp) and GC 42.50% both effectively EXACT; contigs 57 vs 59 within-tol; longest contig differs (430,808 vs 313,663) because the illumina-utils Minoche pre-filter would not run (cutadapt-on-raw substituted) -> more contiguous assembly. Stretch (ANI vs PS5, CheckM completeness/contamination) computing. Out of scope: korormicin NMR/HR-MS chemistry, ddPCR copy numbers (wet-lab), IMG/JGI gene count 4429 (external service), MEGAX ML tree (GUI). No fabrication concern.

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

Whether Pseudoalteromonas sp. strain McH1-7, isolated from disease-resistant Montastraea cavernosa corals, can act as a probiotic that chemically and genomically produces antibacterial compounds capable of treating and preventing stony coral tissue loss disease (SCTLD).

Core claims
  • McH1-7 has broad-spectrum antibacterial activity against SCTLD-associated bacterial isolates and other Gram-positive/Gram-negative targets finding
  • McH1-7 produces the antibacterial compound korormicin, identified by NMR and HRESI/APCIMS finding
  • McH1-7 produces tetrabromopyrrole (TBP), confirmed by LC-HRMS comparison to a standard finding
  • McH1-7 genome encodes marinocine (lodA/lodB L-amino acid oxidase genes) and pseudoalterin-like antibacterial metalloproteases mechanism
  • A previously unknown korormicin (kor) biosynthetic gene cluster (52 kb, 36 ORFs) was identified in McH1-7 and related Pseudoalteromonas strains finding
  • In live coral trials, McH1-7 arrested or slowed SCTLD progression on 68.2% of treated diseased fragments and prevented disease transmission in 100% of cases finding
  • Disease-resistant coral fragments yielded a much higher proportion of culturable antibacterial-producing bacterial isolates than diseased fragments finding
  • McH1-7 is chemically and genomically the best-characterized coral probiotic and a potential alternative to antibiotic treatment for SCTLD resource
Experimental setups
Assay System Perturbation Readout Platform
Plate counts and isolate screening M. cavernosa coral tissue fragments (diseased and disease-resistant) none CFU/ml and % isolates with antibacterial activity SWA and TCBS agar
Disk diffusion assay Live McH1-7 cells vs. target bacterial strains (McT4-15, McT4-56, OfT6-21, plus 32 additional targets) McH1-7 spotting zone of inhibition (ZOI, mm)
Disk diffusion assay McH1-7 organic (ethyl acetate/methanol) extract vs. target bacterial strains chemical extract zone of inhibition (ZOI, mm)
NMR (COSY, HSQC, HMBC, NOESY) and HRESI/APCIMS structural elucidation Purified compound from McH1-7 extract none molecular formula and structure identification (korormicin) HRESI/APCIMS, NMR spectrometer
LC-HRMS McH1-7 crude extract none retention time and mass spectrum match to korormicin and TBP standards LC-HRMS
Whole-genome sequencing, assembly and annotation Pseudoalteromonas sp. McH1-7 none gene content, GC%, biosynthetic gene clusters (antiSMASH), COG assignment MiGA Online; antiSMASH
Comparative genomics (average nucleotide identity) McH1-7 vs. Pseudoalteromonas sp. PS5, J010, P. peptidolytica DSM 14001 none ANI % and gene cluster similarity
Aquarium coral disease trial Live M. cavernosa coral fragments with SCTLD McH1-7 treatment (direct treatment and prophylactic exposure) disease progression arrest/slowing and transmission prevention
Key results
  • McH1-7 arrested or slowed SCTLD progression on treated diseased coral fragments 68.2% (n=22)
  • McH1-7 prevented disease transmission to healthy fragments 100% (n=12)
  • McH1-7 produced a zone of inhibition against nearly all additional target bacteria tested 31 of 32 targets inhibited
  • Pure compound isolated from McH1-7 extract identified as korormicin C25H39NO5, m/z 456.2813 [M+Na]+
  • Tetrabromopyrrole detected in McH1-7 extracts matching standard retention time/isotope pattern
  • Disease-resistant coral fragments (McH1, McH4) had far more antibacterial-producing isolates than diseased fragments (McH2, McH3) 60.36% and 14.41% vs. 0.90% and 0%
  • kor gene cluster identified encoding korormicin biosynthesis 52 kb, 36 ORFs
  • McH1-7 genome highly similar to Pseudoalteromonas sp. PS5 98.91% ANI
Key statistics
  • mean 2.08 ± 0.02 mm (smallest ZOI, McH1-7 vs. V. coralliilyticus OfT6-21)
  • mean 4.50 ± 0.30 mm (largest ZOI, McH1-7 vs. Alteromonas sp. McT4-15)
  • count 68.2% (diseased fragments with arrested/slowed disease after McH1-7 treatment)
  • count 100% (disease transmission prevented by McH1-7)
  • count 67 of 111 (60.36%) (antibacterial-active isolates from disease-resistant fragment McH1)
  • count 16 of 111 (14.41%) (antibacterial-active isolates from disease-resistant fragment McH4)
  • other C25H39NO5; m/z 456.2813 [M+Na]+ (molecular formula/mass of isolated korormicin compound)
  • other 98.91% ANI (average nucleotide identity between McH1-7 and Pseudoalteromonas sp. PS5)

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 excerpt describes a multidisciplinary, largely descriptive study combining in vitro disk-diffusion antibacterial assays, coral tank trials, chemical structure elucidation, and comparative genomics. Quantitative results are reported as simple summary statistics (means with a plus/minus dispersion value, medians and ranges via boxplots) and as raw percentages of coral fragments responding to treatment, rather than through named inferential statistical tests. No hypothesis tests, p-values, or statistical software are explicitly named in the provided text.

Replicationmixed Sample sizeSpecific counts are given for some comparisons (e.g., n = 22 diseased coral fragments treated, n = 12 fragments in transmission-prevention trials, three replicates for zone-of-inhibition assays), but no power analysis or a priori sample-size justification is described. GroupsMcH1-7 activity vs. multiple target bacterial strains; live-cell vs. organic-extract assays; disease-resistant vs. disease-susceptible coral fragments (isolate yield); treated vs. untreated/exposed coral fragments for disease progression and transmission. Pairingunclear Randomization/blindingnot stated Dispersionmixed
Approaches that could also have been used
  • Zone-of-inhibition averages are reported with a plus/minus value without specifying whether it represents SD or SEM.
    Could also: Explicitly labeling the dispersion measure (SD, SEM, or a 95% CI) would also convey the same information — Naming the measure removes ambiguity about whether the spread reflects variability among replicates (SD) or precision of the mean estimate (SEM), which readers often interpret differently, especially with small replicate numbers like three.
  • Disease treatment and transmission-prevention outcomes are summarized as percentages (e.g., 68.2% of n = 22 fragments, 100% of n = 12 fragments) without an accompanying formal statistical test.
    Could also: A Fisher's exact test or chi-square test comparing treated vs. untreated/control fragment outcomes could also be applied — These are binary, categorical outcomes with modest sample sizes, and Fisher's exact test is a standard approach for testing association in such small contingency tables while also yielding a p-value for the observed difference.
  • Percentages of treatment success (e.g., 68.2%, 100%) are reported as point estimates.
    Could also: Reporting a confidence interval for each proportion (e.g., Wilson or Clopper-Pearson CI) could also be included — A CI would communicate the precision of the estimate given the sample size, which is particularly informative when n is in the range of 12-22.
  • The frequency of antibacterial-producing isolates is compared narratively across fragments with different health states (e.g., 60.36% and 14.41% for disease-resistant fragments vs. 0.90% and 0% for fragments that became infected).
    Could also: A chi-square test, Fisher's exact test, or logistic regression relating isolate antibacterial activity to fragment health state could also be used — A formal association test would provide a p-value or odds ratio quantifying the relationship between health state and probability of yielding an antibacterial-producing isolate, complementing the descriptive percentages.
  • Zone-of-inhibition comparisons across many target strains (Fig. 1d) are presented as boxplots based on three replicates per target without a stated formal test between targets.
    Could also: A one-way ANOVA (or Kruskal-Wallis for non-normal/small-sample data) with a post-hoc multiple-comparisons correction could also be applied across target strains — This would allow formal statistical comparison of inhibition potency between targets while accounting for the number of pairwise comparisons being made across the panel of strains.
  • No power or sample-size justification is described for the coral trial group sizes (n = 22, n = 12).
    Could also: An a priori power analysis could also be reported to support the chosen sample sizes — Stating the power analysis would clarify the precision the study was designed to detect, which is useful context for interpreting the treatment and prevention percentages.

What was reproduced

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

Scope — pmid-37024599

Paper: Ushijima et al. 2023, Commun Biol 6:248. "Chemical and genomic characterization of a potential probiotic treatment for stony coral tissue loss disease." PMID 37024599 · PMCID PMC10079959 · DOI 10.1038/s42003-023-04590-y.

Repo: https://github.com/meyermicrobiolab/McH1-7_Probiotics_Trials (authors' own R analysis — P16 not needed; this is the authors' code).

Datasets the paper relies on

accession what used for
PRJNA801145 16S rRNA gene amplicon reads (tank experiment) DADA2/phyloseq community analysis (repo probiotics_amplicons_tank.Rmd)
PRJNA639770 McH1-7 Illumina MiSeq genomic reads SPAdes genome assembly + annotation
JABWPQ000000000 McH1-7 assembled genome (GenBank) the deposited assembly itself
(PS5 genome) Pseudoalteromonas sp. PS5 ANI comparison

NOTE: the BRIEF lists sra:PRJNA639770 as "the" data, but that is the genome BioProject. The 16S amplicon reads — the dataset the shipped R repo analyzes — are under PRJNA801145 (stated in the paper's Data Availability). Both are profiled.

IN SCOPE (pipeline-derived, reproducible)

A. 16S community analysis (repo Rmd; pipeline = cutadapt → DADA2 → Silva 138.1 → phyloseq)

  • A-down (deterministic, from shipped tables): re-run the Rmd downstream on the shipped silva_nochloronomito_otu_table.txt (24 samples × 429 ASVs) + taxa table → reproduce abundance-filter ASV/genera counts (ps1/ps2/ps5/ps10), CLR-PCA, PERMANOVA dist.clr ~ coral*date (Aitchison), betadisper distance-to-centroid. Deterministic → expect EXACT.
  • A-up (full pipeline, from raw reads PRJNA801145): cutadapt-trimmed paired reads (3 MiSeq runs: NS1968R/NS1995/NS2051) → DADA2 filterAndTrim(truncLen=150,150, maxEE=2,2, truncQ=2, maxN=0, rm.phix) per run → mergeSequenceTables → removeBimeraDenovo → assignTaxonomy(silva_nr99_v138.1) → remove chloroplast/mitochondria/Eukaryota/unclassified. Regenerates 447 raw → 429 filtered ASVs + per-sample read-tracking (dada_read_stats{1,2,3}.txt). Param-sensitive but truncLen fully specified → expect exact/within-tol.

B. Genome assembly (PRJNA639770; pipeline = illumina-utils → cutadapt → SPAdes 3.13.0)

  • contigs >1000 bp (reported 59), total assembled length (5,134,549 bp), longest contig (313,663 bp), G+C (42.5%).
  • Stretch: MiGA/CheckM completeness/contamination (reported 100% / 1.9%); ANI to PS5 (reported 98.91%); antiSMASH BGCs.

OUT OF SCOPE (wet-lab / manual / external — not attempted)

  • Chemical characterization of korormicin (NMR, HR-MS, structure) — wet-lab.
  • ddPCR of Kor23 gene copy number (KOR.R only plots lab-measured copy numbers; the measurement is wet-lab) — the plot is trivially re-drawable but the data are not pipeline-derived.
  • Antibiotic / infection inhibition assays, coral fragment trials — wet-lab.
  • Gene count = 4429 comes from the IMG/JGI annotation pipeline (IMG Genome ID 2881214049), an external online service; a local annotator (Prokka/Bakta) gives a different number → reported as a soft target, not a 1:1 claim.
  • MEGAX ClustalW + ML tree of korormicin enzymes — GUI/manual.

Reproduction order

  1. A-down (deterministic R) — quick floor, exact.
  2. B (SPAdes assembly) — concrete genome stats.
  3. A-up (DADA2 from raw reads) — full upstream pipeline.
  4. Stretch: ANI, MiGA/CheckM, antiSMASH.
Figures / tables: table
A1
Reported
24 samples
Reproduced
24
exact
A2
Reported
447 DADA2 chimera-removed ASVs
Reproduced
447 (faithful merge replicating authors' code); 1003 (corrected merge)
exact
A3
Reported
429 ASVs after tax filter
Reproduced
429 (shipped table AND full faithful pipeline)
exact
A4
Reported
263 genera
Reproduced
263 (shipped table); 260 (full faithful DADA2)
exact
A5
Reported
ps1 148/104
Reproduced
148/104
exact
A6
Reported
ps2 72/57
Reproduced
72/57
exact
A7
Reported
ps5 21/16
Reproduced
21/16
exact
A8
Reported
ps10 9/6
Reproduced
9/6
exact
A-track
Reported
DADA2 read tracking (3 runs)
Reproduced
byte-identical to shipped dada_read_stats
exact
A9
Reported
PERMANOVA coral*date R2
Reproduced
coral 0.3107 / date 0.1352 / coral:date 0.5541 (saturated model)
partial
A10
Reported
betadisper figure
Reproduced
ANOVA F=0.308 p=0.90; per-date means reproduced
partial
B1
Reported
59 contigs >1000 bp
Reproduced
57
within tolerance
B2
Reported
5,134,549 bp total
Reproduced
5,134,546 bp (3 bp diff)
within tolerance
B3
Reported
313,663 bp longest
Reproduced
430,808 bp (more contiguous; Minoche pre-filter not reproduced)
did not match
B4
Reported
42.5% G+C
Reproduced
42.50%
exact
B6
Reported
MiGA 100% / 1.9%
Reproduced
running (CheckM, 2259849)
m.public.grade.pending
B7
Reported
ANI vs PS5 98.91%
Reproduced
running (fastANI+enveomics, 2259849)
m.public.grade.pending

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 85/100

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: 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 +2

Strong partial reproduction with no fabrication concern. Input data (shipped 16S cutadapt reads, SRA WGS PRJNA639770) is fully available and used 1:1, and nearly every reported number reproduces exactly or near-exactly: byte-identical DADA2 read-tracking, exact ASV/genera filter cascade (447→429→263, ps1/2/5/10), genome length within 3 bp, and GC 42.50% exact. The two notable items are audit findings on the authors' side that do not break their numbers: the published ASV counts only reproduce when the authors' DADA2 merge bug is replicated (corrected merge gives ~2× ASVs), and the longest-contig differs (430,808 vs 313,663 bp) because the un-reproducible Minoche pre-filter was substituted — both explainable, with identical total length/GC confirming the same underlying assembly. Overall judgement: solid, explainable deviations rather than a clean 1:1 or a substantive discrepancy.

🤝
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.

64.8 k
tokens (I/O) · 3.5 M incl. cache
11 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.