Single-cell RNA-sequencing highlights a curtailed NK cell function in convalescent COVID-19 pregnant women.
The main results reproduced: recomputed values matched the published ones within tolerance.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓Same input data as the authors
- ✓Reported values were directly comparable
- ✓The central claim held under reproduction
- 🟡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
- 🟡Overall, the reproduction showed a material discrepancy
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, and reproduced 1:1 on the headline + structural results. The Zenodo deposit ships a single processed Seurat v5 object (obj.Rds, md5 8bfdd79f... = matches the Zenodo record). The only 'code' link is upstream CellRanger (no FASTQ shipped) so the in-scope reproduction is the downstream pipeline output from obj.Rds, run with Seurat 5.5.0 on «our HPC» SLURM. EXACT: 30,394 cells, 8 sample libraries + 2 groups, and 25 clusters all match the shipped object exactly. HEADLINE biological claim ('curtailed NK function' = cytotoxic-gene downregulation): I identified NK clusters independently by canonical markers (CD3D<1 & CD3E<2 & NKG7>10 -> cluster 7 = CD16+ NK, cluster 18 = CD56bright NK) since no cell-type annotation is shipped, then ran Seurat DE PregRCOVID vs PregHC: all 7 named genes are downregulated and 6/7 are significant (padj<0.05); GZMH is downregulated but not significant. Result is robust to the NK-set choice (adding the CD3-intermediate cluster 22 barely changes it). NOT ATTEMPTED / NOT REPRODUCIBLE: (a) the exact NK I/II/III subtype naming and their individual up/down proportion directions - the cluster->cell-type mapping is NOT in the object (hard ~20%); overall NK fraction did decrease but I could not reproduce a NK subset increasing; (b) CellRanger FASTQ->counts (raw reads not deposited); (c) flow-cytometry validation (wet-lab); (d) the 55,588/46,594 pre-final-filter counts (object is already post-filter). No fabrication concerns: every reported structural number is exactly recoverable from the shipped data, and the headline DE direction reproduces under an independent NK definition.
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.
-
v1 current initial assessment Score 87assessed: 2026-06-14 ⛓ d39bec0de3f7
✎ 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.
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
- 2026-06-14
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15no 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: sonnetWhether SARS-CoV-2 infection 'rewires' the maternal immune response during active infection and whether it 'resets' back to normal during the convalescent/recovery phase in pregnant women.
- ★ NK cells show a sustained reduction during active SARS-CoV-2 infection and persisting after recovery in pregnant women finding
- ★ scRNA-seq shows SARS-CoV-2 infection rewires gene expression profiles of NK cells, monocytes, CD4+ and CD8+ effector T cells, and antibody-producing B cells in convalescent pregnant women finding
- ★ Gene pathways for cytotoxic function, type I & II interferon signalling, and pro-/anti-inflammatory responses in NK and CD8+ cytotoxic T cells are attenuated in recovered pregnant women compared with healthy pregnancies finding
- ★ NK cells from convalescent pregnant women show diminished levels of the cytotoxic proteins perforin, CD122 and granzyme B, validating the scRNA-seq findings finding
- Two independent geographical cohorts (Malaysia discovery, Germany validation) were immunophenotyped using a 14-colour flow cytometry panel method
- scRNA-seq was performed using a Chromium Single Cell 3' Gel Bead Chip and Library Kit (10x Genomics, Drop-seq method) method
- ★ SARS-CoV-2 infection deranges the adaptive immune response in pregnant women even after recovery and may contribute to post-COVID-19 sequelae mechanism
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| 14-color flow cytometry immunophenotyping | PBMCs from pregnant women (Preg-HC, Preg-INF, Preg-R; Malaysia and Germany cohorts) | SARS-CoV-2 infection / convalescence | proportions of immune subsets (NK, CD4+, CD8+, monocytes, B cells, NKT, Tregs) and CD8+ T cell differentiation subsets | BD LSRFortessa Cell Analyzer; FlowJo analysis |
| single-cell RNA sequencing (scRNA-seq) | PBMCs from convalescent pregnant women vs healthy pregnancies | SARS-CoV-2 infection (convalescent phase) | gene expression profiles and pathway activity in NK, monocytes, CD4+, CD8+ effector T cells, and B cells | 10x Genomics Chromium Single Cell 3' Gel Bead Chip and Library Kit |
| 13-plex bead-based cytokine/chemokine assay | serum from pregnant women (validation cohort) | SARS-CoV-2 infection/recovery | concentrations (pg/mL) of IL-1β, IFN-α2, IFN-γ, TNF-α, MCP-1, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-18, IL-23, IL-33 | LEGENDplex Human Inflammation Panel-1; BD LSRFortessa Cell Analyzer |
| 3-plex bead-based serological IgG assay | serum from Preg-R and Preg-HC women (validation cohort) | SARS-CoV-2 infection/recovery | IgG antibody levels against Spike S1, Nucleocapsid, and Spike RBD | LEGENDplex SARS-CoV-2 Serological IgG Panel |
| flow cytometry (surface and intracellular staining) | NK cells from PBMCs of convalescent pregnant women | SARS-CoV-2 infection/recovery | cytotoxic protein levels (perforin, CD122, granzyme B) | BD LSRFortessa Cell Analyzer |
- ▼ NK cell frequency is sustainedly reduced during active infection and after recovery from SARS-CoV-2 in pregnant women
- – scRNA-seq reveals rewired gene expression profiles in NK, monocytes, CD4+, CD8+ effector T cells and B cells of convalescent pregnant women
- ▼ Cytotoxic function, type I/II interferon signalling, and pro-/anti-inflammatory pathways are attenuated in NK and CD8+ T cells of recovered vs healthy pregnant women
- ▼ Perforin, CD122 and granzyme B protein levels are diminished in NK cells of convalescent pregnant women
- – Proportions of naïve, intermediate effector memory (EM II), and late effector memory (EM III) CD8+ T cell subsets differ significantly among Preg-HC, Preg-INF, and Preg-R groups p ≤0.05
- count N = 19 (Preg-HC=11, Preg-INF=4, Preg-R=4) (Cohort 1 (Malaysia, discovery) composition)
- count Healthy controls N=20; recovered N=34 (cytokine/antibody); recovered N=14 (immunophenotyping) (Cohort 2 (Germany, validation) composition)
- pvalue p ≤0.05 (significance threshold for Kruskal-Wallis test with Dunn's multiple comparisons across Preg-HC, Preg-INF, Preg-R groups)
- other >7.1 million deaths; >778 million infected (as of April 2025) (global SARS-CoV-2 pandemic burden, background statistic)
- other 10%-30% (proportion of individuals developing post-COVID-19/Long COVID condition following infection during pregnancy)
- other almost 6 in 100 individuals (estimated proportion suffering Long COVID/post-COVID-19 syndrome among ~778 million reported cases)
- count 1.0 × 10^6 live cells for 14-color FACS panel; 0.5 × 10^6 cells for scRNA-seq (cell input amounts per assay from thawed PBMCs)
- count 100,000-200,000 cells acquired per sample (flow cytometry acquisition per PBMC sample)
Statistical methods review
Model: sonnetA 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 used two independent geographical cohorts (discovery: Malaysia, N=19; validation: Germany, N up to 54) of pregnant women stratified into healthy controls, SARS-CoV-2 infected, and recovered groups. Immune profiling combined 14-colour flow cytometry with UMAP-based unsupervised clustering, scRNA-seq (10x Genomics Chromium), multiplexed cytokine bead assays, and serological IgG quantification. Group comparisons for flow cytometry data were performed with the Kruskal–Wallis test followed by Dunn's post-hoc correction, and results were displayed as violin plots with individual data points; statistical methods for scRNA-seq differential expression are not described in the provided text excerpt.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Kruskal–Wallis test with Dunn's multiple comparisons correction | Percentages of CD8+ T cell subsets (Naïve, intermediate EM II, late EM III) across Preg-HC, Preg-INF, and Preg-R groups (Figure 1h) | Preg-HC N=11, Preg-INF N=4, Preg-R N=4 (Cohort 1 discovery cohort) | not stated |
| UMAP dimensionality reduction with unsupervised clustering (FlowJo) | 14-colour flow cytometry data for immune cell subset identification (Figures 1b–1f) | — | na |
| scRNA-seq differential expression analysis (specific test not stated in provided text) | Gene expression comparisons across NK cells, monocytes, CD4+, CD8+ T cells, B cells between groups | — | not stated |
-
Group comparisons used the Kruskal–Wallis test with Dunn's post-hoc correction across three groups↳ Could also: A one-way ANOVA with Tukey's HSD post-hoc test could also be used if data distribution and variance homogeneity were verified — When normality holds, the parametric ANOVA framework has greater statistical power; Tukey's HSD provides explicit control of the family-wise error rate across all pairwise contrasts, making the correction scope transparent
-
Discovery cohort group sizes are very small (Preg-INF and Preg-R each N=4), yet the same parametric-style threshold (p ≤ 0.05) is applied↳ Could also: Permutation-based or exact tests (e.g., exact Kruskal–Wallis, or pairwise Mann–Whitney U with Bonferroni adjustment) could also be used at such small n — Exact methods do not rely on large-sample asymptotic approximations, which can be unreliable when group sizes fall below ~5; reporting this explicitly helps readers gauge the reliability of the p-values
-
Violin plots are used to display flow cytometry percentage data, and individual dots represent samples↳ Could also: Dot plots overlaid with a median-and-IQR bar, or strip plots with a box plot, could also be used for groups of n=4–11 — Violin plots estimate a density kernel, which can suggest smoother distributions than warranted at n=4; a box-and-dot overlay directly shows each value and the median without distributional assumptions, which is often preferred for very small n
-
Demographic continuous variables (age, gestational age) are summarized as mean ± SD↳ Could also: Median with IQR or range could also be reported, particularly for small, potentially non-normally distributed samples — With group sizes of 4–14, the mean and SD can be heavily influenced by a single outlier; median and IQR are robust summaries and align with the non-parametric inferential tests used for the primary outcomes
-
The specific differential expression method for scRNA-seq is not described in the provided text↳ Could also: Standard scRNA-seq workflows commonly use Seurat with Wilcoxon rank-sum tests or DESeq2 (pseudo-bulk) with negative-binomial Wald tests, with Benjamini–Hochberg FDR correction — Pseudo-bulk DESeq2 aggregates within-donor counts before testing, which better accounts for within-donor correlation and reduces inflation of significant genes compared to cell-level testing; explicitly stating the method and FDR threshold aids reproducibility
-
The two-cohort replication design (Malaysia discovery, Germany validation) was used to support generalizability↳ Could also: A formal meta-analytic or pooled mixed-effects model with cohort as a random or fixed effect could also be applied to jointly estimate effect sizes across cohorts — Pooling cohorts in a mixed model would increase power, allow direct estimation of between-cohort heterogeneity, and yield a single effect-size estimate with a confidence interval rather than requiring qualitative agreement between two separate analyses
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.
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.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-40661959
Paper: Salker et al. 2025, Front Immunol — "Single-cell RNA-sequencing highlights a curtailed NK cell function in convalescent COVID-19 pregnant women." DOI 10.3389/fimmu.2025.1560391.
Code link: github.com/10XGenomics/cellranger (third-party upstream aligner — NOT
authors' own analysis code). Data: Zenodo 10.5281/zenodo.14066080 = a single file
obj.Rds (406 MB, md5 8bfdd79ff35de4a01eec37377ccfb2a1) = the processed Seurat v5
object (30,394 cells, normalized, integrated, clustered).
What the shipped artifact actually contains
obj.Rds is a post-QC, CCA-integrated, clustered Seurat object. meta.data columns:
orig.ident, nCount_RNA, nFeature_RNA, Patient (PregHC/PregRCOVID), Type (a1..a4,b5..b8), mitoPercent, percent.ribo, RNA_snn_res.0.8 (20), seurat_clusters (25), cca_clusters (25).
There is NO cell-type-name column — clusters are numeric only; the "NK I/II/III",
"T cell subset", etc. labels from the figures are NOT shipped.
In scope (pipeline-derived, reproducible from obj.Rds)
- Final cell count (30,394 high-quality cells) — directly counted.
- Sample / group structure (8 libraries; 2 groups HC vs post-COVID) — from metadata.
- Number of clusters / immune subtypes (25) — from
cca_clusters. - NK cytotoxic-gene downregulation in Preg-R (PRF1, GZMA, GZMB, GZMH, KLRD1, NKG7, IRF1) — NK clusters identified by canonical markers, then Seurat DE PregRCOVID vs PregHC. This is the paper's HEADLINE biological claim ("curtailed NK function").
- NK proportion shift — fraction of NK cells per group (partial: see below).
Out of scope (not attempted, with reason)
- CellRanger FASTQ→counts — raw FASTQ are NOT in the Zenodo deposit (only the processed object). The upstream alignment cannot be re-run; it is also not the pipeline step the biological claims rest on.
- Exact NK I / NK II / NK III subtype naming + their individual proportion directions — the cluster→cell-type annotation is NOT shipped in obj.Rds, so the authors' exact NK-subset mapping cannot be reproduced 1:1 (the hard ~20%).
- Flow-cytometry validation (perforin, IL-15RB/CD122) — wet-lab, non-pipeline.
- Intermediate counts 55,588 (recovered) and 46,594 (after first filter) — these are pre-final-filter; the shipped object is already post-filter, so they are not derivable from it.
Pipeline used to reproduce
Seurat 5.5.0 / SeuratObject 5.4.0 / R 4.4.3 (conda, on «our HPC» SLURM). Authors used CellRanger v3.0.1 (GRCh38 v3.0.0) + Seurat 5.2.0, standard workflow, CCA integration.
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
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.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
Structural claims (30,394 cells, 8 libraries/2 groups, 25 clusters) reproduce exactly from the md5-verified Zenodo object, and the headline biological claim of curtailed NK cytotoxic function reproduces strongly (7/7 genes down, 6/7 padj<0.05) under an independent NK definition — so the central conclusion holds. The deviations are on the data-availability side, not a defect in the reported values: the authors shipped no cell-type annotation (so the NK I/II/III subset-shift direction, C5, can't be reproduced 1:1) and no pre-filter matrices (so the 55,588/46,594 counts, C6, are uncheckable). Severity is moderate — the secondary NK-subset 'increase' was not observed and GZMH was down but ns (padj=0.97), but magnitude and direction of the core result hold. No fabrication concern.
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-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.