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

eVITTA: a web-based visualization and inference toolbox for transcriptome analysis.

Nucleic Acids Res · 2021
L1 82/100 PQI 94
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)
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)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score -4
✓ 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
  • Overall, the reproduction was clean
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
  • 🟡A deviation was attributed to the published material
How its reproducibility compares
82/100
Reproducibility score
0.4 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 60% of all assessed papers rank 459 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

eVITTA is a web/Shiny TOOL paper; the only pipeline-derived numbers come from its Figure 2 evaluation case study reanalyzing GEO GSE152075 (430 SARS-CoV-2+ vs 54 uninfected) via the easyGEO module. Read the repo source (easyGEO/server/DE.R @110b4d3e): easyGEO's default DE is limma-voom (edgeR TMM -> voom -> lmFit -> eBayes -> topTable). Ran that exact pipeline on the paper's own author-submitted raw counts on «our HPC», design ~sequencing_batch+positivity (paper's stated 'batch effect adjustment'). Result: the Fig 2D anchor genes reproduce 1:1 within tolerance on logFC and raw p-value -- CNGA4 -0.852/0.162 vs reported -0.84/0.169; GNAL -1.632/0.0135 vs reported -1.65/0.0130 -- and all 7 named upregulation markers (IFIT1/2/3, RSAD2, CXCL9/10/11) come out upregulated as reported. Only padj drifts (CNGA4 0.626 vs 0.488) because BH adjustment depends on the exact tested-gene universe, which eVITTA's interactive filter sets marginally differently. A negative control (no-batch design) lands far from the paper, independently confirming batch adjustment is the correct design and that the reported values are genuinely data-derived (no fabrication signal). NOT attempted (out of scope, 80/20): the easyGSEA GSEA olfactory-transduction enrichment (Fig 2B/2C) -- gene-set DB version + GSEA seed/params unpinned; and all UI/feature-demo figures (Fig 1,3,4) which are not pipeline numbers. Verdict: described well enough to reproduce the DE step; result is a faithful 1:1 on logFC+pval for the pinnable claims. All grades provisional for human audit.

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 82
    assessed: 2026-06-15 ⛓ ce6c39068bc5
✎ 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

Can a web-based, end-to-end toolbox (eVITTA) streamline transcriptome analysis—covering retrieval/analysis of NCBI GEO studies, functional profiling, and multi-dataset comparison—to capture both broad and granular gene expression changes in human and model organism data?

Core claims
  • eVITTA is a web-based toolbox with three modules (easyGEO, easyGSEA, easyVizR) that analyze, interpret, and compare transcriptome data from published NCBI GEO studies or new experiments resource
  • easyGEO systematically extracts and processes NCBI GEO data and performs differential expression analysis using edgeR and limma method
  • easyGSEA provides functional profiling supporting both ORA and pre-ranked GSEA with up-to-date, species-specific, combinatorial gene-set databases and rich visualizations method
  • easyVizR provides an integrated intersection-analysis workflow to compare and visualize multiple transcriptome datasets method
  • In SARS-CoV-2 infected human nasopharyngeal transcriptomes, eVITTA recapitulated an interferon-driven antiviral response and additionally identified downregulation of olfactory signal transducers, consistent with anosmia in COVID-19 finding
  • In C. elegans, SAM deficiency (sams-1 RNAi) and sbp-1 RNAi activate innate immune responses, including TLR signaling, in the absence of pathogen infection finding
  • eVITTA pipeline is computationally robust, producing strongly correlated GSEA results across studies differing in platform (microarray vs RNA-seq) and upstream processing finding
  • eVITTA's interactive, user-friendly interface makes transcriptome analysis accessible to both wet and dry lab biologists with multiple entry/exit points resource
Experimental setups
Assay System Perturbation Readout Platform
bulk RNA-seq (reanalysis, DE + GSEA) human nasopharyngeal swab samples SARS-CoV-2 infection vs uninfected differential gene expression and gene-set enrichment (e.g. olfactory transduction) NCBI GEO GSE152075; easyGEO (edgeR/limma), easyGSEA
microarray (reanalysis, GSEA) C. elegans whole worms sams-1 RNAi gene-set enrichment scores / regulated gene sets NCBI GEO GSE70692 (dataset A)
microarray (reanalysis, GSEA) C. elegans whole worms sams-1 RNAi gene-set enrichment scores / regulated gene sets NCBI GEO GSE70693 (dataset B)
bulk RNA-seq (reanalysis, GSEA) C. elegans whole worms sams-1 RNAi gene-set enrichment scores / regulated gene sets NCBI GEO GSE121508 (dataset C)
microarray (reanalysis, GSEA) C. elegans whole worms sbp-1 RNAi gene-set enrichment scores (immune/TLR signaling) NCBI GEO GSE70692 (dataset D)
Key results
  • SARS-CoV-2 infection upregulated antiviral factors (IFIT1/2/3/6, RSAD2) and chemokines (CXCL9/10/11) in nasopharynx
  • Olfactory transduction gene sets were downregulated during SARS-CoV-2 infection
  • Olfactory transducer CNGA4 showed reduced expression in infected NP samples logFC = -0.84
  • Olfactory transducer GNAL showed reduced expression in infected NP samples logFC = -1.65
  • Strong correlation of significantly regulated gene sets between sams-1 datasets A vs B and A vs C R2 = 0.95 and 0.87
  • Strong Spearman correlation of overall enrichment profiles between unfiltered sams-1 datasets A vs B and A vs C rho = 0.77 and 0.72
  • Strong immune signature confirmed in both sams-1 and sbp-1 deficiency, including TLR signaling changes
Key statistics
  • correlation R2 = 0.95 (GSEA ES correlation between sams-1 datasets A and B (Figure 3B))
  • correlation R2 = 0.87 (GSEA ES correlation between sams-1 datasets A and C (Figure 3C))
  • correlation rho = 0.77; pval < 2.2e-16 (Spearman rank correlation, unfiltered datasets A vs B (Figure 3D))
  • correlation rho = 0.72; pval < 2.2e-16 (Spearman rank correlation, unfiltered datasets A vs C (Figure 3E))
  • fold_change logFC = -0.84; pval = 1.69E-01; padj = 4.88E-01 (CNGA4 expression in SARS-CoV-2 infected NP samples)
  • fold_change logFC = -1.65; pval = 1.30E-02; padj = 3.2E-01 (GNAL expression in SARS-CoV-2 infected NP samples)
  • count 430 infected / 54 uninfected (GSE152075 nasopharyngeal swab samples reanalyzed)
  • count 144751 data series, 4.2 million samples (size of NCBI GEO repository as of 23 February 2021)

Statistical methods review

Model: opus

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 bioinformatics web-server (tools) paper rather than a hypothesis-testing study; its statistical content lies in the analytical pipelines it implements and in two re-analysis evaluation studies of published transcriptome datasets. Differential expression is performed with edgeR (TMM normalization) and the limma-voom/lmFit/eBayes framework, functional profiling uses overrepresentation analysis (ORA) and pre-ranked Gene Set Enrichment Analysis (GSEA via fgsea), and multi-dataset comparisons report intersections plus correlation metrics. Results are reported as log2 fold changes with P-values and adjusted P-values (FDR), and concordance between datasets is summarized with R-squared and Spearman correlation.

Replicationbiological Sample sizeSample sizes are those of the re-analyzed public datasets (e.g. 430 infected vs 54 controls for GSE152075); no power/sample-size calculation is described as this is a tool-evaluation paper Groupsinfected vs uninfected (human); RNAi-treated (sams-1, sbp-1) vs control, and cross-platform dataset comparisons (C. elegans) Pairingunpaired Randomization/blindingna Dispersionnone Exact p-valuesyes Effect sizesyes Confidence intervalsno Multiplicity correctionFDR / adjusted P-value (padj) reported alongside raw P-values; specific FDR algorithm not named in main text (limma/fgsea default Benjamini-Hochberg)
Statistical tests used
Test Applied to n Assumptions
limma linear model (weighted least squares lmFit) with eBayes empirical Bayes moderated statistics for differential expression easyGEO DE analysis; SARS-CoV-2 NP swab DE (GSE152075) and C. elegans datasets 430 SARS-CoV-2-infected vs 54 uninfected controls (GSE152075); per-gene replicates per dataset otherwise not enumerated not stated
edgeR TMM normalization (library-size adjustment prior to limma-voom) RNA-seq datasets in easyGEO pipeline not stated
pre-ranked Gene Set Enrichment Analysis (GSEA via fgsea, permutation-based) easyGSEA functional profiling; olfactory transduction gene sets in SARS-CoV-2 data; C. elegans sams-1/sbp-1 comparisons permutation = 1000; min GS size = 15, max GS size = 200 na
overrepresentation analysis (ORA, hypergeometric-style) via gprofiler2 easyGSEA ORA module on gene lists na
coefficient of determination (R-squared) for enrichment-score concordance Figure 3B (datasets A vs B, n = 320, R2 = 0.95) and 3C (A vs C, n = 359, R2 = 0.87) n = 320; n = 359 (gene sets significant in both datasets) na
Spearman rank correlation Figure 3D (A vs B, n = 912, rho = 0.77, pval < 2.2e-16) and 3E (A vs C, n = 961, rho = 0.72, pval < 2.2e-16) n = 912; n = 961 (ranked gene sets) na
Approaches that could also have been used
  • RNA-seq differential expression is performed via the limma-voom/eBayes workflow after edgeR TMM normalization.
    Could also: A count-based negative-binomial framework such as DESeq2 or edgeR's quasi-likelihood (glmQLFit) test could also be applied to the same count data. — Negative-binomial GLM approaches model count-level mean-variance directly and are a widely used alternative; offering or comparing both can help users see how method choice affects gene-level calls.
  • Dataset concordance is summarized with R-squared and Spearman's rho.
    Could also: Reporting these correlation estimates with 95% confidence intervals (e.g. via bootstrap or Fisher transformation) could also be done. — Confidence intervals convey the precision of a correlation in addition to its point estimate, which complements the n and P-value already shown.
  • Significance filtering uses combined raw-P and adjusted-P thresholds that vary by figure (e.g. padj < 0.25, padj < 0.3, padj < 0.25).
    Could also: A single pre-specified FDR cutoff applied uniformly across comparisons could also be used. — A fixed threshold makes the family-wise/false-discovery control consistent across analyses and can make cross-figure comparisons easier to interpret.
  • The FDR/adjusted-P method is referenced generically as 'padj' without naming the procedure in the main text.
    Could also: Explicitly stating the correction method (e.g. Benjamini-Hochberg) and its scope alongside each test could also be provided. — Naming the multiplicity method makes the analysis fully reproducible and helps readers understand exactly which family of tests the correction covered.
  • Pre-ranked GSEA is run with default permutation = 1000 and gene-set size bounds of 15-200.
    Could also: Sensitivity analyses across additional permutation counts or alternative gene-set size limits could also be reported. — Showing robustness to these parameters illustrates how stable enrichment calls are, which is useful documentation for a general-purpose tool.
  • Gene-level expression is displayed using box and violin plots without a stated central-tendency/dispersion summary statistic.
    Could also: Accompanying these plots with mean or median plus SD, IQR, or a 95% CI could also be reported. — An explicit dispersion summary conveys spread numerically and can be helpful for small-n groups in addition to the distributional plot.
Software: R 4.0.2 · edgeR · limma (voom, lmFit, eBayes) · fgsea · gprofiler2 · GEOquery · R Shiny Server 1.5.14.948

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
98
Impact: high
Foundation confidence
Built on 1 assessed reference(s) · mean reproducibility 56/100
partly built on non-reproducible work
Topics

Assessed papers, coloured by verdict. Click a node to open it.

Built on (assessed references) (1)
Cited by (assessed papers) (0)
  • 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.

GSE152075 GEO in Figure (http://semanticscience.org/resource/SIO_000080)
also used by 2 papers:
GSE121508 GEO in Figure (http://semanticscience.org/resource/SIO_000080)
no other assessed paper uses this yet
GSE70692 GEO in Figure (http://semanticscience.org/resource/SIO_000080)
no other assessed paper uses this yet
GSE70693 GEO in Figure (http://semanticscience.org/resource/SIO_000080)
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-34019643 (eVITTA)

Paper: Cheng et al. 2021, eVITTA: a web-based visualization and inference toolbox for transcriptome analysis, Nucleic Acids Res, PMID 34019643 / PMC8218201. Tool type: web application (R/Shiny). Three modules: easyGEO (GEO retrieval + DE), easyGSEA (gene-set enrichment), easyVizR (multi-list comparison).

Nature of the paper

This is a software/tool paper. Most figures (Fig 1, 3, 4, UI panels) are demonstrations of the web interface and are out of scope (manual/UI, no pinnable pipeline number). The one section that produces concrete, pipeline-derived numbers from public data is the Figure 2 evaluation case study reanalyzing GEO GSE152075 (SARS-CoV-2 nasopharyngeal RNA-seq, 430 infected vs 54 uninfected).

In scope (pipeline-derived, attempted)

The easyGEO differential-expression step of the Fig 2 case study. From eVITTA source (easyGEO/server/DE.R, commit 110b4d3e) the default DE method is limma-voom: edgeR DGEListfilter_genes (≥1 CPM in min-group-size samples) → calcNormFactors (TMM) → voomlmFiteBayestopTable, with output columns renamed logFC / pval (P.Value) / padj (adj.P.Val). Paper note confirms "edgeR and limma-voom normalization with batch effect adjustment".

Pinnable claims (Fig 2D, exact):

  • CNGA4: logFC = -0.84, pval = 1.69E-01, padj = 4.88E-01
  • GNAL: logFC = -1.65, pval = 1.30E-02, padj = 3.2E-01

Pinnable claims (Fig 2A / Results, qualitative-directional):

  • Up in infected: IFIT1/2/3/6, RSAD2, CXCL9/10/11 (interferon antiviral response)
  • Fig 2A: 15 most up / 15 most down genes by logFC (volcano)

Out of scope (not attempted, why)

  • easyGSEA GSEA olfactory-transduction enrichment (Fig 2B/2C, Suppl Table S3): depends on a specific gene-set DB version + GSEA seed/parameters not pinned in the paper; secondary to the DE step. Optional last-20% — skipped per 80/20 rule.
  • All UI/feature-demo figures (Fig 1, 3, 4): not pipeline numbers.

Reproduction approach

Run the underlying pipeline (limma-voom) that eVITTA's easyGEO wraps, on the paper's own data (GSE152075 author-submitted raw counts + series metadata), matching eVITTA's defaults and the paper's "batch adjustment" design (~ sequencing_batch + positivity). Compare regenerated logFC/pval/padj for CNGA4 & GNAL and the up-regulated marker directions to the Fig 2 claims. Per brief P16, running this third-party pipeline on the paper's data is a valid reproduction.

Figures / tables: Fig 2DFig 2AFig 2BTable
C1
Reported
CNGA4 logFC = -0.84
Reproduced
logFC = -0.852
within tolerance
C2
Reported
CNGA4 pval = 1.69E-01
Reproduced
pval = 1.62E-01
within tolerance
C3
Reported
CNGA4 padj = 4.88E-01
Reproduced
padj = 6.26E-01
partial
C4
Reported
GNAL logFC = -1.65
Reproduced
logFC = -1.632
within tolerance
C5
Reported
GNAL pval = 1.30E-02
Reproduced
pval = 1.35E-02
within tolerance
C6
Reported
GNAL padj = 3.2E-01
Reproduced
padj = 3.65E-01
within tolerance
C7
Reported
Upregulated IFIT1/2/3/6, RSAD2, CXCL9/10/11
Reproduced
all upregulated (logFC +1.71 to +2.83)
exact

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 82/100

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)
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)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score -4

This is a tool paper; the only pipeline-derived numbers are the Fig 2 GSE152075 SARS-CoV-2 DE case study, and they reproduce faithfully: CNGA4 (-0.852/0.162 vs -0.84/0.169) and GNAL (-1.632/0.0135 vs -1.65/0.0130) match within ~1-4% on logFC and raw p-value, and all 7 antiviral/chemokine markers come out upregulated as reported. The only deviation is in adjusted p-values (CNGA4 0.626 vs 0.488), which is on our side — a marginally different tested-gene universe in the BH correction from eVITTA's unpinned interactive filter — not an authors' defect. The data is public and identical, a negative control confirms the batch-adjusted design and rules out fabrication, and the central conclusion holds. The GSEA enrichment (C8) was left out of scope, so overall this is a solid near-1:1 reproduction.

🤝
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.

116.3 k
tokens (I/O) · 9 M incl. cache
19 min
runtime · 0.02 CPU-h
1.9 GB
peak RAM
1
HPC jobs
hummel
machine