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

Modulation of HERV Expression by Four Different Encephalitic Arboviruses during Infection of Human Primary Astrocytes.

Viruses · 2022
L1 56/100 3/4
⚑ Flagged for review — a reproduced result did not match the reported value

Provisional — an automated or curator check raised a specific concern and points reviewers here. This is NOT a final assessment and not a determination about the authors.

Why this verdict

The main results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
  • 🟡A deviation was attributed to the published material
  • 🟡Reported values were not (fully) derivable from the shared data
  • 🟡The deviation was non-trivial in magnitude
  • 🟡The central claim did not (fully) hold under reproduction
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
56/100
Reproducibility score
1.0 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 16% of all assessed papers rank 979 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; PARTIAL match overall. Third-party-tool reproduction (P16): re-ran the paper's stated pipeline (Trimmomatic -> bowtie2 --very-sensitive-local -k100 on UCSC hg38 -> Telescope 1.0.3 with HERV_rmsk.hg38.v2 transcripts.gtf, 14968 loci -> DESeq2 per virus vs Mock, CPM>0.5, DEH=padj<0.1 & |log2FC|>1) on the paper's own public data SRA PRJNA662366 (18 PE RNA-seq runs; 6 Mock + 3 each ChikV/MayV/OroV/ZikV), all on «our HPC»/SLURM. WHAT REPRODUCED (within-tol): the flagship per-locus fold-change for HERV4_4q22.1 matches all four viruses within Δ<=0.28 log2 (ours 4.389/5.048/5.943/4.806 vs 4.520/5.181/5.665/4.734) with even-stronger significance; 14 of the 15 specifically named common-up HERVs are recovered (recall 0.93, only HML6_19p13.2c missing); MayV&ChikV is the largest pairwise overlap (qualitative claim holds); ZikV induces far fewer HERVs than the other three; all 18 samples align ~99%. WHAT DIFFERED (mismatch): absolute DE-set sizes are ~2.5-3x larger than reported (common-up 48 vs 15; MayV&ChikV 1303 vs 512) -- a consistent single-direction divergence most plausibly from unshrunken MLE log2FC + a CPM/independent-filtering choice the Methods do not fully specify. NO fabrication concern: every reported value is derivable from the shipped data via the described pipeline; high-confidence loci agree, only borderline low-count loci inflate counts. NOT ATTEMPTED: wet-lab (virus prod/titration, astrocyte isolation, infection, RNA-seq generation) and downstream biological interpretation (GO/nearby-gene enrichment), per scope.md.

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 50
    assessed: 2026-06-19 ⛓ 14d868614dd6
✎ 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 2026-06-19
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 four encephalitis-associated arboviruses (Zika, Mayaro, Oropouche, Chikungunya) modulate human endogenous retrovirus (HERV) expression in infected human primary astrocytes, and whether this modulation influences transcription of nearby genes.

Core claims
  • The four arboviruses commonly induce upregulation of HERVs in human primary astrocytes, with 15 HERVs co-modulated by all four viruses, including highly upregulated HERV4_4q22.1 finding
  • Alphaviruses ChikV and MayV show more similar HERV modulation patterns (both upregulated and downregulated families) to each other than to OroV or ZikV, suggesting conserved viral-family-specific regulation finding
  • Upregulated HERVs are concentrated near differentially expressed genes (mostly within 10-50 kb), consistent with a transcription interference mechanism by which HERV activation regulates nearby gene expression mechanism
  • Protein-protein interaction network analysis shows 93 genes co-upregulated between MayV and ChikV, and 14 genes co-upregulated among ChikV, MayV and OroV, related to cell replication, cytoskeleton, vesicle traffic and antiviral response finding
  • A user-friendly web database/interface was built to allow query of differentially expressed HERVs and nearby genes with GO/KEGG/Reactome enrichment resource
  • HERV differential expression was estimated from RNA-seq using Trimmomatic, bowtie2 alignment to hg38, Telescope quantification, and DESeq2 statistical testing (adjusted p<0.1, |log2FC|>1) method
Experimental setups
Assay System Perturbation Readout Platform
bulk RNA-seq human primary astrocytes (from temporal lobectomy tissue) Chikungunya virus (ChikV) infection, MOI 1 differential HERV and gene expression NextSeq 550 (NextSeq 500/550 High Output v2 Kit, 150 cycles)
bulk RNA-seq human primary astrocytes Mayaro virus (MayV) infection, MOI 1 differential HERV and gene expression NextSeq 550
bulk RNA-seq human primary astrocytes Oropouche virus (OroV) infection, MOI 1 differential HERV and gene expression NextSeq 550
bulk RNA-seq human primary astrocytes Zika virus (ZikV) infection, MOI 1 differential HERV and gene expression (analyzed at 48 hpi) NextSeq 550
flow cytometry human primary astrocytes arbovirus infection (ChikV/MayV/OroV/ZikV) percentage of infected cells (confirmation of infection)
immunofluorescence human primary astrocytes arbovirus infection (ChikV/MayV/OroV/ZikV) viral protein detection / infectivity
protein-protein interaction network analysis differentially expressed genes near DEHERVs from astrocyte RNA-seq (ChikV, MayV, OroV) none (computational/bioinformatic analysis) network connectivity, hub genes, betweenness/eigenvector centrality Cytoscape 3.8.0 with BioGRID PPI database
Key results
  • 15 HERVs were commonly upregulated across all four arbovirus infections
  • HERV4_4q22.1 was highly upregulated at similar levels across all four viruses
  • 512 HERV elements were co-modulated by MayV and ChikV, far more than shared with other virus pairs 512 elements
  • MayV and ChikV upregulated the same HERV families (HERVW, HARLEQUIN, PRIMAX, PABLA, MER41, HML4) in similar proportions; HUERSP3 and HUERSP8 were unique to ChikV upregulation
  • Upregulated HERVs are located up to 50 kb from nearby DEGs (majority within 10 kb), with some more distant correlations up to 260 kb (ChikV) and 320 kb (MayV) up to 320 kb
  • 93 genes were commonly upregulated between MayV and ChikV networks, including replication, immune/antiviral response, cytoskeleton and ubiquitination genes 93 genes
  • 14 genes were commonly upregulated among ChikV, MayV and OroV, including replication and antiviral response genes 14 genes
  • ZikV showed a lower number of significantly modulated HERVs at the analyzed timepoint (48 hpi), attributed to its slower replication rate in astrocytes compared to alphaviruses
Key statistics
  • pvalue adjusted p < 0.1 (threshold for calling a HERV differentially expressed (DEH), via DESeq2)
  • fold_change |log2(fold-change)| > 1 (threshold combined with adjusted p-value for DEH calling)
  • count 15 HERVs (HERVs co-modulated (upregulated) by all four arboviruses)
  • count 512 elements (HERVs co-modulated by MayV and ChikV)
  • count 93 genes (genes common to MayV and ChikV PPI networks)
  • count 14 genes (genes common to ChikV, MayV and OroV PPI networks)
  • other up to 260 kb (ChikV) / 320 kb (MayV) (maximum distance of correlated upregulated HERV-gene pairs)
  • count 0.5 CPM (minimum expression threshold for retaining HERVs in analysis)

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 paper re-analyzed RNA-seq data from human primary astrocytes infected with four arboviruses (ChikV, MayV, OroV, ZikV) to quantify differential expression of human endogenous retroviruses (HERVs) using Telescope for quantification and DESeq2 for differential expression testing, applying an adjusted p-value and fold-change threshold to call elements differentially expressed. Results were further explored by correlating differentially expressed HERVs with nearby differentially expressed genes (from a prior published dataset) based on genomic distance, followed by Gene Ontology/KEGG/Reactome enrichment and protein-protein interaction network construction in Cytoscape using BioGRID interaction data.

Replicationunclear Sample sizeRNA-seq was performed in '2 × 9 samples' using two different donors; the exact number of biological replicates per virus/timepoint condition is not explicitly broken down in the text provided GroupsInfected (ChikV, MayV, OroV, ZikV) vs. non-infected astrocyte controls, at virus-specific hours post-infection Pairingunclear Randomization/blindingnot stated Dispersionnone Exact p-valuesno Effect sizesyes Confidence intervalsno Multiplicity correctionNot explicitly named in the text (described only as 'adjusted p-values'); DESeq2's default adjustment method is Benjamini-Hochberg FDR
Statistical tests used
Test Applied to n Assumptions
DESeq2 (Wald test, as implemented in R/Bioconductor DESeq2) Differential expression of HERVs for each virus (ChikV, MayV, OroV, ZikV) vs. non-infected controls RNA-seq performed on 2 × 9 samples; HERVs below 0.5 CPM were filtered before testing not stated
Approaches that could also have been used
  • Differential expression significance was defined using a combined adjusted p-value (< 0.1) and fold-change (|log2FC| > 1) threshold from DESeq2's Wald test.
    Could also: A likelihood ratio test (LRT) within DESeq2, or edgeR/limma-voom pipelines, could also be used for count-based differential expression testing — Different differential expression frameworks (DESeq2 Wald vs. LRT, edgeR, limma-voom) make different distributional assumptions and can be compared to assess robustness of the DEH calls, particularly useful when sample sizes are modest, as is common in RNA-seq of primary cell infections
  • The paper used an adjusted p-value threshold of 0.1 rather than the more conventional 0.05 to call HERVs differentially expressed.
    Could also: A stricter adjusted p-value cutoff (e.g., 0.05) or reporting a range of thresholds with the corresponding number of DEHs at each could also be presented — Showing sensitivity of the DEH count to the chosen adjusted p-value cutoff can help readers gauge how threshold-dependent the reported overlaps (e.g., the 15 shared HERVs) are
  • The multiple-testing correction method for the DESeq2 analysis was not explicitly named in the text, though DESeq2 uses Benjamini-Hochberg FDR by default.
    Could also: Explicitly stating the correction method (e.g., Benjamini-Hochberg) and reporting q-values or adjusted p-values directly in supplementary tables could also be included — Making the correction method explicit in the text, rather than relying on the software default, aids reproducibility and helps readers apply comparable thresholds when reanalyzing or comparing to other studies
  • Effect sizes were reported as log2 fold-change per HERV/gene, without accompanying measures of uncertainty such as standard error or confidence intervals for these estimates.
    Could also: Reporting DESeq2's shrunken log2FC estimates along with their standard errors or 95% confidence intervals (e.g., via the lfcShrink function) could also be presented — Confidence intervals around fold-change estimates convey the precision of each estimate, which is particularly informative when comparing effect magnitudes across the four different viruses
  • Overlap of upregulated HERVs and genes across the four viruses was described using counts and Venn-style comparisons (e.g., 15 shared HERVs, 93 shared genes between MayV and ChikV).
    Could also: A formal statistical test of overlap significance (e.g., hypergeometric test or Fisher's exact test for gene-set overlap) could also be applied — Testing whether the observed overlaps exceed what would be expected by chance, given the total number of DEHs/DEGs per virus, would add a quantitative measure to the qualitative similarity observed between viruses
  • Protein-protein interaction networks were built and described qualitatively (e.g., node centrality, connectivity, common genes) without formal statistical testing of network properties.
    Could also: Network topology could also be evaluated against randomized/permuted networks (e.g., via permutation testing of centrality measures or modularity) to assess statistical significance of observed network features — Comparing observed network metrics to a null distribution generated from randomized networks can help quantify whether the connectivity patterns are distinguishable from chance given the network size and degree distribution
Software: Trimmomatic · Bowtie2 · Telescope · R/Bioconductor DESeq2 · PostgreSQL (relational database for results storage) · Cytoscape 3.8.0 · BioGRID database 4.3.194

What was reproduced

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

Scope — pmid-36423114

Paper: Castro et al. 2022, Viruses 14(11):2505. "Modulation of HERV Expression by Four Different Encephalitic Arboviruses during Infection of Human Primary Astrocytes." PMID 36423114 / PMC9694637 / DOI 10.3390/v14112505.

Nature of the reproduction (P16 — third-party tool on the paper's data)

This paper reuses a public RNA-seq dataset (SRA PRJNA662366, originally generated by the same group, Souza et al.) and applies an existing third-party bioinformatics stack to it. There is no authors' own analysis repo; the "Code" link is the annotation database for the Telescope tool (mlbendall/telescope_annotation_db). Per BRIEF rule 2 this is an equally valid reproduction: we run the described pipeline (Trimmomatic → bowtie2 → Telescope → DESeq2) on PRJNA662366 with the stated parameters and compare derived numbers.

Pipeline described in Methods (the pipeline to reproduce)

  1. Trimming: Trimmomatic (adapters + low-quality reads).
  2. Alignment: bowtie2, --very-sensitive-local, up to 100 alignments reported (-k 100), genome hg38.
  3. HERV/TE quantification: Telescope (default parameters) with the HERV hg38 GTF from mlbendall/telescope_annotation_db (build HERV_rmsk.hg38.v2, transcripts.gtf).
  4. Filtering: remove HERVs with expression < 0.5 CPM.
  5. Differential expression: DESeq2; DEH = adjusted p < 0.1 AND |log2FC| > 1. Each virus (ChikV, MayV, OroV, ZikV) vs Mock.

In scope (pipeline-derived results we attempt)

id reported result source pipeline
C1 15 HERVs upregulated by all four arboviruses (intersection), named in Table 1 Table 1 / Results full pipeline → DESeq2 intersect
C2 HERV4_4q22.1 log2FC ChikV 4.520 / MayV 5.181 / OroV 5.665 / ZikV 4.734; q ≤ 7.63e-30 Results §3 / Table 1 DESeq2 per-virus
C3 512 elements co-modulated by MayV and ChikV (largest pairwise overlap) Fig 1C / Results DESeq2 per-virus ∩
C4 per-virus DEH counts (up/down), Fig 1A bar chart Fig 1A DESeq2 per-virus
C5 total HERV loci passing CPM>0.5 filter (test universe) Methods/Results Telescope counts → CPM filter

Primary targets: C1, C2, C3 (clearly numeric). C4/C5 are secondary (exact numbers not all printed in text; gradeable as partial where the paper only shows a figure).

Out of scope (not attempted — wet-lab / external / manual)

  • Virus production, titration (PFU), astrocyte isolation/culture, MOI-1 infection, RNA extraction, library prep, sequencing (PRJNA662366 generation) — wet-lab.
  • Gene-neighborhood / "nearby gene" enrichment & GO/pathway interpretation, protein-interaction or biological-narrative claims — downstream interpretation, attempted only if pipeline DEH sets reproduce.
  • Any claim about HERV biology/causation — interpretation, not pipeline output.

Data

  • PRJNA662366 — 18 paired-end bulk RNA-seq runs (SRR12610900–SRR12610917): 6 Mock (MockA1-3, MockB1-3), 3 ChikV, 3 MayV, 3 OroV, 3 ZikV. (matches paper's "2 × 9 samples"). Open access via ENA/SRA.

Compute plan

All heavy steps on «our HPC»/SLURM, inputs on «infra». Download (fastq + hg38 + HERV GTF [git-LFS]) on front1. See reproduction/ scripts.

Figures / tables: TableFig 1CFig 1A
C2
Reported
HERV4_4q22.1 log2FC ChikV 4.520/MayV 5.181/OroV 5.665/ZikV 4.734; q<=7.63e-30
Reproduced
4.389/5.048/5.943/4.806 (all Δ<=0.28); padj 1e-41..1.8e-87
within tolerance
C1_list
Reported
the 15 named common-up HERVs (Table 1)
Reproduced
14/15 recovered (recall 0.93); only HML6_19p13.2c missing
within tolerance
C1
Reported
15 HERVs upregulated by all four arboviruses
Reproduced
48 (superset containing 14/15 of the named set)
did not match
C3
Reported
512 elements co-modulated by MayV and ChikV (largest pairwise overlap)
Reproduced
1303 all-DEH / 971 up-only; MayV&ChikV IS the largest pairwise overlap
did not match
C4
Reported
per-virus DEH counts (Fig 1A bar chart)
Reproduced
ChikV 1547 / MayV 1725 / OroV 1403 / ZikV 112 total DEH; ZikV far fewer
partial
C5
Reported
DE test universe (not stated numerically)
Reproduced
6107 of 13894 telescope loci pass CPM>0.5
partial
DESIGN
Reported
18 samples (2x9): 6 Mock + 3 each ChikV/MayV/OroV/ZikV
Reproduced
18 runs downloaded+aligned (99%+), groups reconstruct exactly
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 56/100

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.

56.5 k
tokens (I/O) · 3.7 M incl. cache
12 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.