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

Exploring the key genetic association between chronic pancreatitis and pancreatic ductal adenocarcinoma through integrated bioinformatics.

Front Genet · 2023
L1 91/100 PQI 97
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 🟡
Minor / cosmetic deviation
+1 pts
From: Q6 · Severity of the deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +3
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • Reported values are derivable from the shared data
  • 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
  • 🟡The deviation was non-trivial in magnitude
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
91/100
Reproducibility score
1.0 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 82% of all assessed papers rank 197 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 the core results 1:1. The repo (KaiGuo2023/KaiGuo-CP-PDAC @85474fc) is a data+results dump (only code = ID.R, a probe->symbol mapper); P16 applies, so we reproduced via standard tools (GEOquery/limma, WGCNA results) on the paper's own GEO data. WGCNA is the strongest result: 10/10 module-trait correlations, p-values, key-module gene counts (406/790/370/2032), and module counts (CP 8, PDAC 22) match the paper EXACTLY against the shipped author artifacts -> reported numbers are fully derivable from shipped data (no fabrication indicator). Independent limma DEG re-run on the two discovery sets: GSE91035 PDAC-vs-normal = 3022 vs reported 3120 (within ~3%, within-tol); the overlapping-DEG set = 78 vs reported 85 (within ~8%, within-tol) and independently recovers ALL SIX reported hub genes plus TF PTF1A. GSE143754 CP-vs-control = 368 vs reported 165 (mismatch, ~2.2x): the paper used GEO2R on Affy HTA-2.0 (GPL17586) whose probeset annotation/collapse differs from our gene_assignment->symbol avereps; same order of magnitude, and downstream overlap is unaffected. NOT attempted (last-20% / out of scope): PPI-network hub-gene selection (interactive STRING/Cytoscape), CEL diagnostic AUC=0.968 on validation cohorts, ssGSEA/ESTIMATE/TIMER immune infiltration, TIDE, miRNA (hsa-miR-198) and TF analyses, GO/KEGG enrichment percentages -- mostly qualitative or reliant on external web tools.

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 91
    assessed: 2026-06-15 ⛓ 9ee782f86387
✎ 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 room) · 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

What are the real hub genes and common molecular mechanisms underlying the transformation of chronic pancreatitis (CP) into pancreatic ductal adenocarcinoma (PDAC)?

Core claims
  • CEL (carboxyl ester lipase) is the real hub gene shared between CP and PDAC and is markedly downregulated in PDAC finding
  • Dysfunctional transport of nutrients and trace elements may be a common pathogenic mechanism of CP and PDAC mechanism
  • WGCNA combined with GEO2R/DEG analysis identifies four trait-associated co-expression modules and six candidate hub genes (ALB, CEL, CELA3B, CTRL, PLA2G1B, SYCN) method
  • hsa-miR-198 may be the key upstream miRNA regulating CEL finding
  • CEL expression is strongly correlated with the tumor immune microenvironment (TIME) in PDAC finding
  • Shared genes are enriched in detoxification of copper ion, fat digestion/absorption, long-chain fatty acid transport, and protein digestion/absorption finding
Experimental setups
Assay System Perturbation Readout Platform
Microarray/RNA-seq gene expression profiling (discovery) Human pancreatic tissue (CP and PDAC patients, GSE143754, GSE91035) none (disease vs control) Gene expression / co-expression modules and DEGs GPL17586 (GSE143754); GPL22763 (GSE91035)
WGCNA co-expression network analysis GSE143754 (CP) and GSE91035 (PDAC) datasets none Module-trait correlations, hub gene connectivity WGCNA R package
Differential expression validation (GEO2R/ACLBI) Human CP and PDAC tissue (GSE101462, GSE151945) none (CP vs PDAC) CEL/hub gene expression differences
Immunohistochemistry expression validation PDAC vs normal pancreatic tissue none CEL protein expression Human Protein Atlas (HPA)
Co-expression / functional enrichment analysis (GO/KEGG) TCGA PDAC transcriptome none Genes correlated with CEL; pathway enrichment LinkedOmics, DAVID, ClueGO
miRNA and transcription factor regulatory network prediction CEL gene (CP/PDAC) none Upstream miRNAs and TFs HMDD, miRWalk, TRRUST
Tumor immune microenvironment / immune infiltration analysis TCGA PDAC RNA-seq (178 patients) CEL high vs low expression Immune/stromal scores, immune cell infiltration, checkpoint genes, ICB response ESTIMATE, ssGSEA, TIMER, TISIDB, TIDE
Key results
  • CEL is significantly differentially expressed between CP and PDAC in both validation sets and markedly low in PDAC
  • Red module positively correlated with CP r=0.83, p=1e-04
  • Cyan module negatively correlated with CP r=-0.81, p=2e-04
  • Black module highly associated with PDAC r=0.84, p=7e-10
  • Light cyan module negatively associated with PDAC r=-0.83, p=2e-09
  • 212 genes shared across highly correlated CP and PDAC modules; 85 overlapping DEGs identified 212 genes; 85 DEGs
  • Detoxification of copper ion accounted for 60% of GO terms among shared genes 60%
  • Patients with CP had a 14-fold increased risk of developing PDAC (background) 14-fold
Key statistics
  • correlation r=0.83, p=1e-04 (Red module association with CP (GSE143754))
  • correlation r=-0.81, p=2e-04 (Cyan module association with CP)
  • correlation r=0.84, p=7e-10 (Black module association with PDAC (GSE91035))
  • correlation r=-0.83, p=2e-09 (Light cyan module association with PDAC)
  • fold_change 14-fold (Increased PDAC risk in CP patients (Lowenfels 1993))
  • count 212 shared genes; 85 overlapping DEGs (Shared genes/DEGs between CP and PDAC modules)
  • count 6 hub genes (ALB, CEL, CELA3B, CTRL, PLA2G1B, SYCN) (Hub genes from WGCNA and PPI intersection)
  • other 60% (Detoxification of copper ion share of GO terms)

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 bioinformatics study integrated public microarray and RNA-seq datasets from GEO to identify shared hub genes between chronic pancreatitis and pancreatic ductal adenocarcinoma. The analytical pipeline combined weighted gene co-expression network analysis (WGCNA) on two discovery cohorts with GEO2R differential expression analysis, followed by PPI network topology scoring, external validation in two additional GEO cohorts and TCGA, and immune microenvironment characterization. Results were reported primarily as module-trait Pearson correlation coefficients, adjusted p-value thresholds, fold changes, and Kaplan-Meier survival curves with log-rank test p-values.

Replicationbiological Sample sizeSample sizes determined by available GEO datasets; the only stated size criterion was a minimum of 10 samples per discovery group for WGCNA; no formal power calculation described GroupsCP vs normal controls; PDAC vs normal controls; CP vs PDAC across multiple independent GEO cohorts and TCGA; CEL-high vs CEL-low PDAC patients Pairingunpaired Randomization/blindingnot stated Dispersionnone Exact p-valuesyes Effect sizesyes Confidence intervalsno Multiplicity correctionAdjusted p-values applied in GEO2R DEG discovery step (specific correction method not stated; GEO2R defaults to Benjamini-Hochberg FDR); raw p < 0.05 used in ACLBI validation analysis; no correction stated for WGCNA module-trait p-values or immune microenvironment comparison tests
Statistical tests used
Test Applied to n Assumptions
Pearson correlation (WGCNA module-trait associations) Module-trait correlation heatmaps for CP (Figure 1G) and PDAC (Figure 1H); threshold r > 0.3 and p < 0.05 n=15 (GSE143754: 6 CP + 9 controls); n=33 (GSE91035: 25 PDAC + 8 controls) not stated
GEO2R differential expression analysis (|log2FC| > 1, adjusted p < 0.05) DEG identification in discovery cohorts GSE143754 and GSE91035 (volcano plots, Figures 2F–G) n=15 (GSE143754); n=33 (GSE91035) not stated
ACLBI-based differential expression analysis (raw p < 0.05, no adjustment stated) Validation DEG analysis comparing CP and PDAC in GSE101462 and GSE151945 n=14 (GSE101462: 10 CP + 4 PDAC); n=6 (GSE151945: 3 CP + 3 PDAC) not stated
Logistic regression Diagnostic performance of CEL discriminating TCGA PDAC from GTEx normal pancreas samples 178 PDAC patients (TCGA); GTEx normal n not stated not stated
Wilcoxon rank sum test Two-sample group comparisons in tumor immune microenvironment analyses TCGA PDAC cohort (n=178 stated); exact per-test n not stated not stated
Kruskal-Wallis test Multi-group comparisons in tumor immune microenvironment analyses TCGA PDAC cohort (n=178 stated); exact per-test n not stated not stated
Log-rank test with Kaplan-Meier survival curves Survival analysis between immune subgroups and clinical characteristics in PDAC TCGA PDAC cohort (n=178 stated) not stated
Approaches that could also have been used
  • Pearson correlation was used for WGCNA module-trait associations with small discovery sample sizes (n=15 and n=33)
    Could also: Spearman rank correlation could also be applied for module-trait associations — Spearman correlation does not assume linearity or normality of the trait variable and is more robust to outliers, which can be relevant when sample sizes are small; it is sometimes used as a sensitivity check alongside Pearson in WGCNA workflows
  • DEGs in discovery cohorts were identified using GEO2R with an adjusted p-value threshold of 0.05, but the specific multiple-testing correction method was not named
    Could also: Explicitly applying and naming Benjamini-Hochberg FDR correction, or using the limma package directly, would also control the false discovery rate with full transparency — Naming the correction method allows readers to reproduce the analysis and to assess the stringency of FDR control; limma with its moderated t-statistics is a purpose-built and widely benchmarked approach for small-n microarray differential expression
  • Hub genes were ranked by seven separate network topology algorithms and the intersection of top-ranked genes was taken via an UpSet plot
    Could also: A rank aggregation method (e.g., Borda count or RobustRankAggreg) or a single well-motivated centrality measure could also summarize the multi-algorithm scores into a single ranked list — Intersection-of-thresholds does not have a statistical null distribution and is sensitive to the choice of cutoff per algorithm; rank aggregation provides a continuous score across all algorithms and a more reproducible selection criterion
  • Immune cell infiltration was estimated using ssGSEA (24 gene sets) and TIMER deconvolution independently
    Could also: CIBERSORT, MCP-counter, or EPIC could also estimate immune cell proportions from bulk RNA-seq data — Different deconvolution algorithms use distinct reference signatures and modeling assumptions; reporting which tool was used for which specific comparison, or comparing agreement across tools, allows readers to contextualize how method choice may influence infiltration estimates
  • Survival associations between CEL expression or immune subgroups and outcome were assessed with Kaplan-Meier curves and log-rank tests
    Could also: Multivariable Cox proportional hazards regression could also assess these survival associations — Cox regression allows adjustment for clinical covariates (e.g., stage, age, grade) that may confound the relationship between immune subgroup or gene expression and survival, and produces hazard ratios with confidence intervals as quantitative effect-size estimates
  • Logistic regression was used to evaluate the diagnostic performance of CEL in distinguishing PDAC from normal pancreas tissue
    Could also: ROC curve analysis with reported AUC and 95% confidence interval could also quantify diagnostic accuracy — AUC provides a threshold-independent summary of discriminatory ability and its confidence interval supports direct comparison with other candidate biomarkers; it is a common complement to logistic regression in biomarker diagnostic studies
Software: R 3.4.1 · WGCNA (R package) · preprocessCore (R package, normalize.quantiles) · estimate (R package) · UpSet (R package) · GEO2R (NCBI online tool) · ACLBI (online normalization and DEG tool) · ClueGO / Cytoscape · Metascape (including MCODE and cytoHubba plugins) · DAVID · LinkedOmics · GEPIA · TIMER · TISIDB · TIDE · Xiantao tool (online)

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
11
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.

Data lineage

The datasets this paper uses (text-mined from the full text via Europe PMC), and which other assessed papers stand on the same data. A shared dataset is a factual link — not a judgement.

GPL17586 GEO in Results (http://purl.org/orb/Results)
also used by 2 papers:
GPL17077 GEO in Results (http://purl.org/orb/Results)
also used by 1 paper:
GPL22763 GEO in Results (http://purl.org/orb/Results)
no other assessed paper uses this yet
GSE101462 GEO in Results (http://purl.org/orb/Results)
no other assessed paper uses this yet
GSE143754 GEO in Results (http://purl.org/orb/Results)
no other assessed paper uses this yet
GSE151945 GEO in Results (http://purl.org/orb/Results)
no other assessed paper uses this yet
GSE91035 GEO in Results (http://purl.org/orb/Results)
no other assessed paper uses this yet
rs488087 RefSNP in Discussion (http://purl.org/orb/Discussion)
no other assessed paper uses this yet

What was reproduced

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

Scope — pmid-37501719

Paper: Guo et al. 2023, Front Genet. "Exploring the key genetic association between chronic pancreatitis (CP) and pancreatic ductal adenocarcinoma (PDAC) through integrated bioinformatics." DOI 10.3389/fgene.2023.1115660.

Repo: github.com/KaiGuo2023/KaiGuo-CP-PDAC @85474fc7 — a data + results dump, not a runnable pipeline. Only one R script (ID.R, probe→symbol mapping for GSE143754). Everything else is shipped intermediate/result files (WGCNA module zips, GEO2R DEG result CSVs, ssGSEA xlsx, TIDE csv). P16 applies: standard third-party tools (GEO2R/limma, WGCNA) on the paper's own GEO data.

In scope (pipeline-derived, attempted)

# Result Pipeline Reproduction strategy
C1 WGCNA module-trait r & p (CP red/cyan, PDAC black/lightcyan) WGCNA Read shipped moduleTraitCor/Pvalue.csv, compare to paper
C2 WGCNA module gene counts (406/790/370/2032) WGCNA Per-module CSV filenames (*_N.csv)
C3 WGCNA module counts (CP 8, PDAC 22) WGCNA Count shipped module folders + grey
C4 DEG counts: GSE143754=165, GSE91035=3120 GEO2R/limma Independent re-run of limma on GEO data
C5 Overlapping DEGs = 85 set intersection From C4 DEG lists

C1–C3 verify the authors' shipped WGCNA artifacts against the paper text (internal-consistency / fabrication check). C4–C5 are an independent re-execution of the front-end DEG pipeline on the raw GEO series matrices (GEOquery + limma, |log2FC|>1 & adj.P<0.05).

Out of scope (not attempted — why)

  • Hub gene selection (ALB/CEL/CELA3B/CTRL/PLA2G1B/SYCN → CEL): depends on a PPI network (STRING/Cytoscape) + WGCNA intersection; manual/interactive, under-specified. (last-20%)
  • CEL AUC=0.968 on validation (GSE101462/GSE151945): pROC on validation data — feasible but downstream of hub selection; deferred (80/20).
  • ssGSEA / ESTIMATE / TIMER immune infiltration, TIDE, miRNA (hsa-miR-198), TFs (STAT5A/B, PTF1A): descriptive/qualitative ("increased", "negatively correlated"), few hard numbers; many rely on external web tools (TIDE web, HMDD, NetworkAnalyst). Out of scope.
  • GO/KEGG enrichment percentages: qualitative term lists, no exact reproducible scalar.
Figures / tables: Fig3Fig2Table1Fig2H
C1a
Reported
CP red module r=0.83, p=1e-04
Reproduced
r=0.8332, p=1.145e-04
exact
C1b
Reported
CP cyan module r=-0.81, p=2e-04
Reproduced
r=-0.8143, p=2.194e-04
exact
C1c
Reported
PDAC black module r=0.84, p=7e-10
Reproduced
r=0.8433, p=7.279e-10
exact
C1d
Reported
PDAC lightcyan module r=-0.83, p=2e-09
Reproduced
r=-0.8340, p=1.665e-09
exact
C2a
Reported
CP red module 406 genes
Reproduced
406
exact
C2b
Reported
CP cyan module 790 genes
Reproduced
790
exact
C2c
Reported
PDAC black module 370 genes
Reproduced
370
exact
C2d
Reported
PDAC lightcyan module 2032 genes
Reproduced
2032
exact
C3a
Reported
CP 8 modules
Reproduced
8 (7 colored + grey)
exact
C3b
Reported
PDAC 22 modules
Reproduced
22 (21 colored + grey)
exact
C4a
Reported
GSE143754 DEGs = 165
Reproduced
368 (probe-level 1223)
did not match
C4b
Reported
GSE91035 DEGs = 3120
Reproduced
3022
within tolerance
C5
Reported
overlapping DEGs = 85
Reproduced
78 (contains all 6 hub genes ALB/CEL/CELA3B/CTRL/PLA2G1B/SYCN + TF PTF1A)
within tolerance

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 91/100

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 🟡
Minor / cosmetic deviation
+1 pts
From: Q6 · Severity of the deviation 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +3

WGCNA (C1–C3) reproduces 1:1 against the authors' shipped artifacts — 10/10 module-trait correlations, p-values, key-module gene counts (406/790/370/2032), and module counts (CP 8 / PDAC 22) all exact, so the reported numbers are demonstrably derivable from the shared data with no fabrication signal. The only material deviation is on the DEG side, and it is on our methodology: GSE143754 yields 368 vs the reported 165 (~2.2x) because the paper used GEO2R defaults on Affy HTA-2.0 while we used an avereps gene-level collapse (GSE91035 3022 vs 3120 and overlap 78 vs 85 are within tolerance). Severity is moderate — same order of magnitude, direction preserved — and the core conclusion holds because the independently derived overlap recovers all 6 reported hub genes plus PTF1A. Overall: a solid, partly-exact reproduction with an explainable preprocessing-driven discrepancy on our side, not the authors'.

🤝
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.

149.3 k
tokens (I/O) · 10.8 M incl. cache
16 min
runtime · 0.02 CPU-h
2 GB
peak RAM
3 (1 failed)
HPC jobs
hummel
machine