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

Antibiotic Drug screening and Image Characterization Toolbox (A.D.I.C.T.): a robust imaging workflow to monitor antibiotic stress response in bacterial cells <i

F1000Res · 2021
L1 88/100 3/4
Why this verdict

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

Reproduced on the brainbox compute brainarbeit.com
✓ 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
  • Reported values are derivable from the shared data
  • Any deviation was negligible
  • The central claim held under reproduction
What did not (or only partly)
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
88/100
Reproducibility score
0.8 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 74% of all assessed papers rank 276 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

A.D.I.C.T. (Mayer et al., F1000Research 2021) is a Fiji/ImageJ + R + Celltool semi-automated imaging workflow for bacterial antibiotic-stress morphology. Code: GitHub Image-processing-and-analysis-workflows/A.D.I.C.T. (commit 5312a66, GPL-3, Zenodo 10.5281/zenodo.5342923); DATA on OSF registration fnj5g (NOT Zenodo). Computed on «our HPC» («job» + 2250020). RESULT: described well enough and reproduces 1:1 on the deterministic pipeline outputs. C1-C4 per-timepoint segmented-cell counts 412/455/338 (total 1205, Fig 7A) reproduce EXACTLY by grouping the deposited merged table Puromycin_stress_CN32.csv. C5 time-dependent area increase reproduces EXACTLY: mean Area 2.019->2.439->2.891 um^2 monotonic; Shapiro p<2.2e-16 (non-normal); pairwise Wilcoxon(holm) Areacondition all pairs p<=4.1e-12, and Areaprotein gives Ffh-FtsY n.s. / FtsY-L1 *** / Ffh-L1 *** EXACTLY matching the significance annotations hard-coded in the repo R script. C6 DBSCAN cluster counts reproduce EXACTLY: L1=6 (eps=30), Ffh=10 (eps=3), FtsY=3 (eps=3), minPts=5 (Fig 8). C7 Celltool PCA (Fig 5) PARTIAL: Mode 1 = 96.36% reproduces the 'Mode 1 explains majority of shape variation' claim, but the reported smallest retained mode 0.8% vs our 0.21% differs (sensitive to the exact multi-step alignment procedure / which model subset Fig 5 used). C8 dataset size PARTIAL: 283 binary/brightfield images observed across the 9 protein x condition groups vs 286 reported (off by 3). NOT attempted (out of scope): the brightfield raw-image -> ROI segmentation end-to-end, which the authors state needs ~1-2 working days of manual ROI correction by a trained person (not deterministically reproducible), and all wet-lab work (strains, microscopy acquisition, viability assay). No fabrication indicators: every reproduced value is directly derivable from the shipped data+code.

💻 Code ↗ 🗄 Data: 10.5281/zenodo.5342923

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 75
    assessed: 2026-06-19 ⛓ 1abd0262da90
✎ 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
2026-06-25
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19
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

The paper tests whether a semi-automated, combined brightfield/fluorescence imaging workflow (A.D.I.C.T.) using ImageJ2/FIJI, Celltool and R can robustly and reproducibly monitor antibiotic-induced morphological and protein-biogenesis stress responses in bacterial cells in vivo.

Core claims
  • A.D.I.C.T. provides a robust, reproducible semi-automated imaging workflow for monitoring antibiotic stress response in bacteria in vivo method
  • 'Percentile' thresholding is more efficient than 'Otsu' for segmenting brightfield images of Shewanella cells, requiring less manual correction finding
  • Puromycin-induced stress causes increased cell area and abnormal curvature over time finding
  • Rifampicin and chloramphenicol induce similar cell shape/area stress responses as puromycin finding
  • Fluorescence intensity of Ffh and FtsY fusion proteins decreases over time under puromycin because translation is inhibited and existing proteins are degraded mechanism
  • Cellular area is the decisive quantitative parameter for monitoring puromycin-induced stress response finding
  • The workflow combines customized ImageJ2/FIJI macros, Celltool, and R packages that can be reproduced and adapted by other users resource
  • Density-based clustering (dbscan) on cellular area and mean gray value can distinguish clusters related to fusion protein amount and cellular area method
Experimental setups
Assay System Perturbation Readout Platform
Brightfield microscopy and binary image processing Shewanella putrefaciens CN-32 recombinant strains (Ffh, FtsY, L1-mVenus fusions) puromycin/rifampicin/chloramphenicol cell segmentation, ROI area Olympus IX 71 (100x/NA 1.49), ImageJ2/FIJI
Live-cell slimfield fluorescence time-lapse microscopy Shewanella putrefaciens CN-32 recombinant strains (Ffh, FtsY, L1-mVenus fusions) puromycin/rifampicin/chloramphenicol fluorescence intensity (mean gray value, integrated density) Andor iXON Ultra EMCCD, Andor Solis
Cell shape analysis and modelling (Celltool) Shewanella putrefaciens CN-32 cells puromycin/rifampicin/chloramphenicol shape modes (PCA), curvature, area Celltool
Viability/colony dilution assay Shewanella putrefaciens CN-32 cells puromycin colony density on LB-agar after serial dilution Fusion-Gel-Illuminator
Statistical hypothesis testing Cell area and fluorescence measurement datasets none normality (Shapiro-Wilk), pairwise significance (Wilcoxon rank sum) RStudio v1.1.463, R 3.6.1
Unsupervised clustering (dbscan) ImageJ2/FIJI results tables (cellular area, mean gray value) puromycin clusters distinguishing fusion protein amount vs cellular area R dbscan package
Key results
  • Puromycin-stressed cells show increasing cell area and increasingly abnormal curvature over the time-course
  • Rifampicin- and chloramphenicol-stressed cells show increased cellular area and curvature over time, similar to puromycin
  • Fluorescence intensity of Ffh and FtsY decreases over time under puromycin treatment
  • L1 fluorescence amount is higher after 30 minutes of puromycin induction, reason unclear
  • Integrated density increases, supporting cellular area as the key indicator of puromycin stress
  • Puromycin-treated cells show drastically reduced colony density versus steady-state in viability/dilution assay
  • 'Percentile' thresholding produces fewer artificial gaps and requires less manual correction than 'Otsu' for brightfield segmentation
  • Cellular area differs significantly between L1, Ffh and FtsY strains overall, but not between Ffh and FtsY specifically, and is strongly influenced by extreme values
Key statistics
  • count 0 min: n=412; 30 min: n=455; 60 min: n=338 (cell counts per timepoint in pooled fusion-protein boxplot (Figure 7A))
  • other 200 µg/ml (puromycin concentration used to induce stress)
  • other 25 µg/ml (rifampicin concentration used for confirmatory stress induction)
  • other 50 µg/ml (chloramphenicol concentration used for confirmatory stress induction)
  • pvalue p<0.05=*, p<0.01=**, p<0.001=***; confidence level 0.95 (Wilcoxon rank sum test significance thresholds for pairwise comparisons)
  • other 0.8% (lower cutoff of variance explained for Celltool shape modes to be considered)
  • other OD600 0.5 (exponential growth phase at which cells were sampled for imaging)
  • count ≥2000 frames, 17.76 ms integration time (slimfield fluorescence time-lapse acquisition parameters)

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 describes an image-analysis workflow (ADICT) combining ImageJ2/FIJI image processing, Celltool-based cell-shape modeling (including PCA), and R-based statistical evaluation of bacterial cell area, mean gray value, and integrated density under antibiotic stress. Normality of measurement distributions was assessed with the Shapiro-Wilk test, and pairwise group comparisons were then made with the non-parametric Wilcoxon rank sum test, with significance reported via threshold-based asterisks (p<0.05/0.01/0.001). An unsupervised dbscan clustering approach was additionally applied to cellular area and fluorescence intensity measurements to explore subpopulation structure.

Replicationmixed Sample sizeCell-level sample sizes are given for pooled fusion-protein data in Figure 7A (n=412, 455, 338 at 0/30/60 min); time-course replicates are described as taken on different days, but a formal biological-replicate count or power calculation is not stated Groupssteady-state vs. antibiotic-treated (puromycin, rifampicin, chloramphenicol) cells across time points, and across protein-fusion strains (Ffh, FtsY, L1) Pairingunpaired Randomization/blindingnot stated DispersionIQR Exact p-valuesno Effect sizesno Confidence intervalsno Multiplicity correctionnone stated
Statistical tests used
Test Applied to n Assumptions
Shapiro-Wilk test assessing normality of distributions of cell area, mean gray value (mgv), and integrated density (IntDen) measurements stated
Wilcoxon rank sum test (non-parametric, pairwise) pairwise comparisons of cellular areas/curvature/fluorescence over time and between strains (Ffh, FtsY, L1) and antibiotic conditions (puromycin, RIF, CM), e.g. Figures 4, 6, 7 for Figure 7A pooled fusion-protein samples: 0 min n=412, 30 min n=455, 60 min n=338; not stated for other comparisons stated (chosen after Shapiro-Wilk normality test)
Principal component analysis (PCA) / shape mode decomposition Celltool-based cell shape modeling across time points and conditions (Figures 5, 6C, 6G) not stated
Density-based spatial clustering (dbscan) unsupervised clustering of cellular areas and mean gray value to distinguish fusion-protein subpopulations not stated
Approaches that could also have been used
  • Many pairwise Wilcoxon rank-sum comparisons are made across time points, strains, and antibiotic conditions without a stated multiple-comparison correction.
    Could also: A Kruskal-Wallis test across all groups followed by a post-hoc test with a correction such as Benjamini-Hochberg FDR or Bonferroni — This would control the family-wise error rate or false discovery rate across the many comparisons being made, which some readers find useful when interpreting cumulative significance across multiple tests.
  • Significance is reported using threshold-based asterisks (p<0.05/0.01/0.001) rather than exact p-values.
    Could also: Reporting exact p-values alongside the significance thresholds — Exact values let readers gauge the strength of evidence directly and support later meta-analyses or re-analysis, complementing the asterisk convention.
  • Group differences are described qualitatively (e.g., 'significant time dependent increase') without an accompanying effect-size statistic.
    Could also: Reporting an effect size such as the Hodges-Lehmann estimator or rank-biserial correlation alongside the Wilcoxon result — An effect size conveys the magnitude of the difference in addition to statistical significance, which can be informative when comparing strains or time points.
  • The text notes that differences between strains 'are strongly influenced by extreme values,' observed via boxplots.
    Could also: A formal outlier-robust summary (e.g., reporting median/IQR explicitly, or a sensitivity analysis excluding flagged extreme values) alongside the existing Wilcoxon test — This can help readers see how much a result depends on a small number of extreme observations, complementing the already non-parametric, rank-based test choice.
  • Time-course replicates are described as collected on different days, and cell-level measurements from these replicates appear to be pooled for comparisons (e.g., n=412/455/338 pooled cells per time point).
    Could also: A mixed-effects or hierarchical model with day/replicate as a random effect — This approach can account for potential clustering of cells measured within the same experimental day, which is a standard way to model nested biological replicate structure when many cells are measured per replicate.
  • Manual correction of binary images (ROI drawing) is part of the image-processing pipeline, and no statement is made about whether this step was performed blind to sample condition.
    Could also: Performing and reporting blinded manual corrections (e.g., file names anonymized to condition) during image curation — Blinding this manual step is a standard way to further guard against unconscious observer influence during image-based quantification, complementing the workflow's existing automation.
Software: ImageJ2/FIJI · Celltool · R (base) 3.6.1 · RStudio 1.1.463 · R package dplyr · R package dbscan

What was reproduced

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

Scope — pmid-35707454 (A.D.I.C.T.)

Paper: Mayer B, Schwan M, Thormann KM, Graumann PL. Antibiotic Drug screening and Image Characterization Toolbox (A.D.I.C.T.): a robust imaging workflow to monitor antibiotic stress response in bacterial cells. F1000Research 2021. PMID 35707454 · PMCID PMC9178280 · DOI 10.12688/f1000research.51868.3

Code: https://github.com/Image-processing-and-analysis-workflows/A.D.I.C.T. (GPL-3.0; archived Zenodo v1.0.4 = 10.5281/zenodo.5342923). Note the repo is described as 100% Jupyter Notebook with module folders: Binary image processing, Celltool script, ROI generator, Projection tool, Intensity measurement tool, R scripts, Viability assay.

Data (IMPORTANT — on OSF, not Zenodo): https://osf.io/ynkz3/ "Test data for A.D.I.C.T. workflow". Zenodo holds the code archive, OSF holds the data.

Pipeline overview

  1. Brightfield → ROI (Fiji macros): 16→8-bit, Kuwahara denoise, percentile threshold, watershed, manual correction (~1–2 working days by a trained person), ROI extraction + area measurement.
  2. Fluorescence → intensity (Fiji macros): annotate/crop after photobleaching, SD projection, background subtraction, Otsu mask, per-cell intensity (Mean/IntDen).
  3. Shape modelling: Celltool (PCA on cell outlines) — Fig 5.
  4. Statistics / clustering (R 3.6.1): dplyr/ggplot2/tidyverse; Shapiro-Wilk + Wilcoxon rank-sum (p<0.05); dbscan density clustering — Fig 8.

IN SCOPE (pipeline-derived, attempted)

  • C1–C4 Per-timepoint and total segmented-cell counts (Fig 7A: 412/455/338, total 1205) — reproduced EXACTLY from deposited merged table 06. tables/merged csv final table/Puromycin_stress_CN32.csv (1205 rows), grouped by condition. DONE.
  • C5 Time-dependent area increase (Fig 7A) — direction reproduced (mean Area 2.019→2.439→2.891 µm²); exact Wilcoxon p-values pending R rerun.
  • C6 DBSCAN cluster counts per strain (Fig 8: L1=6, Ffh=10, FtsY=3) — needs eps/minPts from the repo R script; queued for «our HPC».
  • C7 Celltool PCA Mode-1 dominance (Fig 5, min mode 0.8%) — queued for «our HPC».
  • C8 Dataset image counts (286 brightfield, 9 conditions) — verify on «infra».

OUT OF SCOPE (not attempted, with reason)

  • Brightfield raw-image → ROI segmentation end-to-end: the authors state it requires ~1–2 working days of manual ROI correction by a trained person, so it is not deterministically/automatically reproducible. We instead start from the deposited ROI/measurement tables (downstream of manual curation).
  • Wet-lab: strain construction (Ffh/FtsY/L1 fusions in S. putrefaciens CN-32), microscopy acquisition (Olympus IX71), antibiotic treatments, viability assay.
  • External: no external benchmark/ground-truth dataset is used by the paper.

Reproduction strategy

Inputs are small CSV tables + PNG outlines (OSF, <1 MB for tables). Plan: clone the repo and download OSF data onto «infra» (front1), then run the repo's R scripts + Celltool/dbscan under a small SLURM job; pull back the cluster counts, PCA mode table, and Wilcoxon p-values; compare to Fig 5/7/8. The cell-count claim (C1–C4) needed only the merged table and is already reproduced.

Status of compute

«our HPC»/«infra» VPN tunnel currently DOWN (central). Offline prep complete; repo+OSF download and the R/SLURM run are queued for when «host» ssh recovers.

Figures / tables: Figure 7AFig 7AFigure 8Figure 5
C1
Reported
412 cells at 0 min (Fig 7A)
Reproduced
412
exact
C2
Reported
455 cells at 30 min (Fig 7A)
Reproduced
455
exact
C3
Reported
338 cells at 60 min (Fig 7A)
Reproduced
338
exact
C4
Reported
1205 total cells
Reproduced
1205
exact
C5
Reported
significant time-dependent area increase (Fig 7A)
Reproduced
mean 2.019->2.439->2.891 um^2; Wilcoxon(holm) all condition pairs p<=4.1e-12; protein pairs match repo ***/n.s. annotations exactly
exact
C6
Reported
DBSCAN L1=6, Ffh=10, FtsY=3 (Fig 8)
Reproduced
L1=6, Ffh=10, FtsY=3 (eps L1=30 / Ffh=FtsY=3, minPts=5)
exact
C7
Reported
Celltool PCA Mode 1 dominant, min mode 0.8% (Fig 5)
Reproduced
Mode 1 = 96.36% (dominant=majority of shape variation); 5 modes retained, smallest 0.21% vs reported 0.8%
partial
C8
Reported
286 brightfield images, 9 conditions
Reproduced
283 binary images, 9 protein x condition groups (off by 3)
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 88/100

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.

81 k
tokens (I/O) · 4 M incl. cache
23 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.