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

Single-cell transcriptional dynamics of flavivirus infection.

Elife · 2018
L1 90/100 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)
Minor / cosmetic deviation
+1 pts
From: Q2 · Endpoint comparability 🟡
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 -5
✓ What held up
  • Same input data as the authors
  • 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
  • Overall, the reproduction was clean
What did not (or only partly)
  • 🟡Reported values were only indirectly comparable
How its reproducibility compares
90/100
Reproducibility score
0.9 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 79% of all assessed papers rank 211 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 -> reproduced 1:1. Applied the paper's described method (Spearman of host gene expression vs per-cell intracellular virus abundance) to the paper's own data (GSE110496, 2260 viscRNA-seq cells) on «our HPC». Cohort structure (DENV MOI 0/1/10, ZIKV MOI 0/1, 4/12/24/48h) matches exactly. Central result reproduces 1:1 for dengue: ER-stress/UPR genes are top-positive (DDIT3 r=+0.55, #2 of the transcriptome; HSPA5/SEC61G/SSR3/TRIB3/SEC61B/ATF3 all positive) and cytoskeleton genes are the most anti-correlated (ACTB/TUBB/ACTG1 r approx -0.5), at the |r|>0.3 / top-1% threshold the paper states. ZIKV reproduces the same ER-stress positive signature with smaller magnitude because its viral dynamic range is narrower (3% vs 45% of reads; MOI<=1) -- consistent with the paper. NOT ATTEMPTED (80/20): wet-lab pro/antiviral validation (out of scope), t-SNE/GO-enrichment/100-bootstrap CIs/time-switcher trajectories (hard last 20%), and bit-identical authors'-env reproduction (singlet would not import on modern pandas -> used the identical scipy Spearman it wraps, documented). No fabrication signal; all compared values derive directly from the shipped GEO data.

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 90
    assessed: 2026-06-16 ⛓ eb85ed79e348
✎ 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
2026-06-16
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-16
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 virus-inclusive single-cell RNA-Seq approach (viscRNA-Seq) that simultaneously measures host transcriptome and intracellular viral RNA in the same cell reveal heterogeneity in flavivirus (dengue and Zika) abundance and identify pro- and antiviral host factors and their dynamics during infection?

Core claims
  • viscRNA-Seq simultaneously quantifies whole single-cell host transcriptome and intracellular viral RNA from the same cell, including non-polyadenylated viruses via a virus-specific oligo method
  • Intracellular DENV/ZIKV abundance is extremely heterogeneous across single cells, spanning up to ~1000-fold and up to a quarter of all reads per cell finding
  • Correlating per-cell gene expression with intracellular virus abundance identifies host factors and cellular functions (ER translocon, signal peptide processing, N-linked glycosylation, membrane trafficking) involved in flavivirus replication finding
  • DENV and ZIKV induce partially overlapping but virus-specific host responses, with some genes (e.g. ID2, HSPA5) playing opposite roles in the two infections finding
  • Loss- and gain-of-function screens validated novel proviral (RPL31, TRAM1, TMED2) and antiviral (ID2, CTNNB1) host factors mediating DENV infection finding
  • Several genes (e.g. COPE) switch correlation sign with vRNA over the infection time course, suggesting multiple temporal roles during infection finding
  • ER unfolded protein response/ER stress genes (e.g. DDIT3/CHOP) are upregulated and correlated with virus, while actin/microtubule cytoskeleton genes are anticorrelated, reacting at higher virus thresholds mechanism
  • Cell cycle phase does not appreciably affect intracellular DENV abundance at early time points finding
Experimental setups
Assay System Perturbation Readout Platform
single-cell RNA-Seq (viscRNA-Seq, modified Smart-seq2 with virus-specific oligo + ERCC spike-ins) human hepatoma Huh7 cells DENV serotype 2 strain 16681 infection at MOI 1 and 10 single-cell host mRNA expression and intracellular viral RNA reads per million transcripts over time (4,12,24,48 hr) illumina NextSeq 500, ~400,000 reads per cell
single-cell RNA-Seq (viscRNA-Seq) human hepatoma Huh7 cells ZIKV Puerto Rico strain PRVABC59 infection at MOI 1 single-cell host mRNA expression and intracellular viral RNA abundance illumina NextSeq 500
qPCR virus screening single sorted Huh7 cells (384-well plates) DENV/ZIKV infection intracellular vRNA content / fraction of infected cells
single-cell RNA-Seq replicate (independent smaller-scale time course) human hepatoma Huh7 cells DENV infection (1/5th scale) reproducibility of gene-vRNA correlations
loss-of-function and gain-of-function (knockdown/overexpression) screens human cells (Huh7) knockdown/overexpression of candidate host factors (RPL31, TRAM1, TMED2, ID2, CTNNB1) effect on DENV infection (validation of proviral/antiviral role)
GO enrichment analysis (computational) DENV-infected Huh7 single-cell transcriptomes none enriched pathways among top 1% correlated genes (|r|>0.3) at 4 hr and 48 hr PANTHER online service
Key results
  • Fraction of cells with >10 virus reads increases with MOI and time, saturating at 48 hr post infection
  • Intracellular virus content can increase ~1000-fold with no saturation across infected cells 1000-fold
  • Stress response gene DDIT3/CHOP expression saturates after a ~10-fold increase 10-fold
  • Up to ~25% of reads per cell are vRNA-derived (~10^5 reads) ~25% of reads
  • COPE switches from strongly negative to strongly positive correlation with vRNA as infection proceeds <-0.3 to >+0.3
  • Six genes switch from negative to positive correlation with DENV vRNA and 11 genes switch in the opposite direction 6 and 11 genes
  • In ZIKV high-virus cells CASP3 shows positive correlation while CASP6 shows negative correlation; ATF4 and EDEM1 differ between viruses
  • Positive apoptosis regulator BBC3/PUMA shows only modest correlation with DENV amount r=0.17
Key statistics
  • count ~7500 single cells screened, ~2100 sequenced (total cells across all conditions)
  • count 380 cells screened per condition, ~100 sequenced per condition (per experimental condition)
  • count ~400,000 reads per cell (sequencing depth)
  • correlation |Spearman r| > 0.3 (threshold defining top ~1% correlated/anticorrelated genes)
  • correlation r=0.17 (BBC3/PUMA correlation with intracellular DENV amount)
  • correlation 500 or more virus reads per million transcripts (cell inclusion cutoff for DENV vs ZIKV correlation comparison)
  • other 100 bootstraps over cells (uncertainty estimation for correlation coefficients)

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.

The study uses a single-cell RNA-seq design (viscRNA-Seq) in which host gene expression and intracellular viral RNA are quantified in the same cell, and the main analysis correlates each gene's expression with viral RNA abundance across many single cells using Spearman's rank correlation. Uncertainty on correlation coefficients is estimated by 100 bootstraps over cells, and gene sets are characterized with Gene Ontology enrichment (PANTHER). Supplementary comparisons (e.g., bystander vs control cells) use a nonparametric Kolmogorov-Smirnov test with Bonferroni correction, and replicate consistency is assessed with Pearson's r between experiments. Results are largely reported as correlation coefficients with bootstrap standard deviations rather than as group-mean significance tests.

Replicationbiological Sample size~380 cells screened per condition with ~100 sequenced; ~7500 cells screened and ~2100 sequenced in total; an independent smaller-scale DENV experiment (1/5th cell numbers) was performed for reproducibility; statistical power attributed to large numbers of single-cell replicates GroupsDENV- vs ZIKV-infected vs uninfected Huh7 cells across MOIs (1, 10) and time points (4, 12, 24, 48 hr); within-cell gene–vRNA correlations Pairingunpaired Randomization/blindingnot stated DispersionSD Exact p-valuesyes Effect sizesyes Confidence intervalsno Multiplicity correctionBonferroni (applied to the bystander-cell Kolmogorov-Smirnov tests)
Statistical tests used
Test Applied to n Assumptions
Spearman's rank correlation coefficient gene expression vs intracellular vRNA across single cells (Figure 2A–E, Figure 3, time-resolved analyses) all sequenced/infected single cells (~2100 sequenced overall; subsets such as cells with ≥500 or ≥1000 vRNA reads/million for some panels) stated
Bootstrap resampling (100 bootstraps over cells) uncertainty/error bars on correlation coefficients (Figures 2 and 3) cells per condition not stated
Gene Ontology enrichment analysis (PANTHER) top ~1% correlated/anticorrelated genes; pathway interpretation at 4 hr and 48 hr post-infection gene sets with |correlation| > 0.3 not stated
Kolmogorov-Smirnov test (nonparametric) differential expression between bystander cells and control cells (Figure 2—figure supplement 5) control vs zero-vRNA cells from 4 and 12 hr time points stated
Pearson's r correlation of per-gene correlation coefficients between the large-scale and independent small-scale DENV experiments (Figure 2—figure supplement 1) genes not stated
Least-squares piecewise-linear fit in log-log space inferring reaction thresholds of expression vs vRNA (Figure 2B–C, Figure 2—figure supplements 2–3) na
Approaches that could also have been used
  • Gene–virus associations were quantified with Spearman's rank correlation, chosen to avoid an explicit noise model and to reduce outlier sensitivity.
    Could also: Generalized linear models for count data (e.g., negative-binomial regression as in DESeq2/edgeR) or zero-inflated models that include vRNA level as a covariate. — A model-based approach would additionally provide per-gene effect-size estimates with formal confidence intervals and could adjust for covariates such as sequencing depth or time, complementing the rank-based, assumption-light correlation.
  • Uncertainty on the genome-wide correlations is summarized by a |r|>0.3 magnitude threshold together with bootstrap standard deviations.
    Could also: A genome-wide multiple-testing framework such as Benjamini-Hochberg FDR (or permutation-based null distributions) applied to per-gene correlation p-values. — An FDR-controlled gene list would give an explicit expected false-positive rate across the ~20,000 tested genes alongside the effect-size threshold already used.
  • Dispersion is reported as the standard deviation of 100 bootstrap replicates of each correlation coefficient.
    Could also: Reporting bootstrap 95% confidence intervals (e.g., percentile or BCa intervals) for the coefficients. — A confidence interval conveys the plausible range of each estimate directly and is often preferred for communicating precision, especially when bootstrap distributions are asymmetric.
  • The bystander-vs-control comparison used Kolmogorov-Smirnov tests with Bonferroni correction.
    Could also: A Benjamini-Hochberg FDR correction, or a count-based differential-expression model across the same gene family. — FDR control is typically more sensitive than Bonferroni for transcriptome-wide screens, which could increase power to detect subtle bystander effects while still bounding false discoveries.
  • Reproducibility was assessed by comparing correlation coefficients between a large-scale and an independent smaller-scale DENV experiment using Pearson's r.
    Could also: Concordance metrics such as Lin's concordance correlation coefficient, Spearman's rank correlation, or Bland-Altman agreement analysis between replicates. — Concordance or agreement measures capture both correlation and systematic offset/scale differences between replicates, adding to the linear-association summary provided by Pearson's r.
  • Thresholds in the expression–vRNA relationship were inferred with least-squares piecewise-linear fits in log-log space.
    Could also: Formal changepoint/segmented regression with estimated breakpoint confidence intervals, or smooth nonlinear fits (e.g., GAMs). — These approaches would provide statistical uncertainty on the inferred threshold location and a basis for comparing fit quality across genes, supplementing the descriptive piecewise fits.
Software: PANTHER (online GO enrichment service) · Smart-seq2 single-cell RNA-seq protocol (library preparation) · illumina NextSeq 500 (sequencing platform)

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
180
Impact: high
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.

What was reproduced

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

Scope — pmid-29451494 (Zanini et al. 2018, eLife, flavivirus single-cell)

Paper: Single-cell transcriptional dynamics of flavivirus infection. Zanini F, Pu SY, Bekerman E, Einav S, Quake SR. eLife 2018;7:e32942. PMC5826272. Method (viscRNA-Seq): virus-inclusive single-cell RNA-seq of Huh7 cells infected with dengue (DENV2) or Zika (ZIKV), Smart-seq2-style, plate-based. Data: GEO GSE110496 — GSE110496_RAW.tar = 2260 per-cell count TSVs (feature<TAB>count, ~60721 Ensembl gene rows + HTSeq __* QC rows). Per-cell metadata (virus, moi, time[h], n_<virus>_molecules) in GSE110496_series_matrix.txt.gz. Tool: github.com/iosonofabio/singlet (authors' generic scRNA package; HEAD 66d3b13581f571450c9551e99d86845cce992d32). It is a framework, not a paper-specific script; its correlation routine is a thin wrapper over scipy Spearman. Per brief rule P16, applying the described method/tool to the paper's own data is valid.

In scope (pipeline-derived, reproduced)

  1. Cohort structure — # cells sequenced; split by virus, MOI, timepoint. Pipeline: parse GSE110496 series matrix + count files.
  2. Intracellular virus abundance per cell — vRNA reads per million transcripts; dynamic range / max fraction of reads. Pipeline: n_virus_molecules / total reads.
  3. Spearman correlation of each host gene vs virus abundance (the paper's central analysis, Fig 2–4). Reproduced for dengue and zika; checked the named ER-stress/UPR (positive) and cytoskeleton (negative) genes + global top/bottom.

Out of scope (not attempted, and why)

  • Wet-lab validation of pro/antiviral host factors (siRNA/CRISPR) — experimental.
  • t-SNE/clustering figures, GO enrichment, 100-bootstrap CIs, time-switcher gene trajectories (Fig 2G/4C) — the hard last ~20%; the qualitative + magnitude claims above already establish a clear 1:1 on the core method. Not attempted (80/20).
  • Bit-identical reproduction of authors' singlet version (pinned env not published); we record our resolved env instead.
Figures / tables: Fig 2
C1
Reported
~2100 cells sequenced
Reproduced
2260 cells in GSE110496
within tolerance
C2
Reported
DENV: 4 uninfected/4 MOI1/4 MOI10 plates
Reproduced
dengue moi0=335, moi1=487, moi10=534
exact
C3
Reported
ZIKV: 4 uninfected/4 MOI1 plates
Reproduced
zika moi0=410, moi1=494
exact
C4
Reported
timepoints 4/12/24/48 h
Reproduced
all four present for both viruses
exact
C5
Reported
DENV vRNA up to ~1/4 of reads, ~1e5 reads
Reproduced
max fraction 0.449, 252272 molecules, 448875 ppm
within tolerance
C6
Reported
Spearman rank correlation gene~virus abundance
Reproduced
Spearman (scipy 1.17 rankdata+Pearson)
exact
C7
Reported
|r|>0.3 = top ~1% of transcriptome
Reproduced
key |r|>0.3 genes in top ~0.5%
within tolerance
C8
Reported
DENV: ER-stress/UPR genes (DDIT3,ATF3,TRIB3,SEC61B/G,SSR1/2/3,HSPA5) up with virus
Reproduced
DDIT3 +0.55(#2), HSPA5 +0.50, SEC61G +0.47, SSR3 +0.43, TRIB3 +0.43, SEC61B +0.41, ATF3 +0.37, SSR1 +0.33, SSR2 +0.27
within tolerance
C9
Reported
DENV: cytoskeleton (ACTB,ACTG1,TUBB) down with virus
Reproduced
ACTB -0.56, TUBB -0.52, ACTG1 -0.50 (most anti-correlated genes)
within tolerance
C10
Reported
ER-stress signature positive in BOTH viruses
Reproduced
zika DDIT3 +0.47, TRIB3 +0.45, ATF3 +0.43
within tolerance
C11
Reported
HSPA5 upregulation dengue-specific
Reproduced
dengue HSPA5 +0.50 vs zika +0.01
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 90/100

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: Q2 · Endpoint comparability 🟡
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 -5

This is a strong reproduction run on the authors' own deposited data (GSE110496, 2260 cells) with the authors' own described method (Spearman of host gene expression vs per-cell virus abundance). The central conclusion reproduces 1:1: ER-stress/UPR genes top-positive (DDIT3 r=+0.55), cytoskeleton most anti-correlated (ACTB r=−0.56), ER-stress positive in both viruses, HSPA5 dengue-specific (+0.50 vs +0.01). The only frictions are paper-side approximations (it reports ~2100, "a quarter of reads", and qualitative gene lists rather than exact per-gene r — hence q2 yellow) and a benign environment workaround (scipy Spearman replacing the non-importable singlet, fully documented). No fabrication signal; all values derive from the shared data.

🤝
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.

120.5 k
tokens (I/O) · 6.3 M incl. cache
18 min
runtime · 0.04 CPU-h
4.7 GB
peak RAM
3
HPC jobs
hummel
machine