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

Dissection of multiple sclerosis genetics identifies B and CD4+ T cells as driver cell subsets.

Genome Biol · 2022
L1 No data access 2/4
Why this verdict

Part of the results reproduced; minor but material deviations remained.

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: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
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 🟡
Input / endpoint not comparable 1:1
+1 pts
From: Q1 · Data identity 🔴
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +7
✓ What held up
  • Any deviation was negligible
What did not (or only partly)
  • 🔴Could not use the authors’ exact input data
  • 🔴Reported values were only indirectly comparable
  • 🟡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 central claim did not (fully) hold under reproduction
  • 🟡Overall, the reproduction showed a material discrepancy
No data access Data access not granted

This paper has a computational component, but its primary data is legally or ethically access-restricted — identifiable patient cohorts, rare-disease genomes, or controlled-access biobanks that cannot be openly shared. The reproduction therefore could not be attempted. That is a neutral verdict: it does not mean the result is wrong or that the authors fell short — only that, for legitimate privacy reasons, it cannot be independently checked from public data. We deliberately do NOT assign a 0–100 score here, because a low number would wrongly read as a failed reproduction.

Reproduction agent’s raw note

DROP - data_restricted. The paper's headline computational results are cell-type-specific stratified LD-score regression (S-LDSC, bulik/ldsc) of MS GWAS heritability partitioned over open-chromatin annotations built from GSE74912 (Corces hematopoiesis ATAC-seq). The method (LDSC) is public and the ATAC half (GSE74912) is public and resolves (130 samples). BUT the dependent variable for every gradeable claim - the IMSGC-2019 MS GWAS summary statistics (GWAS Catalog study GCST009597; 8,278,136 variants; 14,802 cases / 26,703 controls) - is controlled-access: GWAS Catalog reports fullPvalueSet=false (no hosted sumstats) and imsgc.net/downloads is a request-only form gated by a Data Access Committee with no-redistribution terms; the paper itself states the sumstats are 'available via request to the IMSGC'. No reported number is derivable from the public ATAC data alone (the paper gives no standalone per-cell-type peak count). Therefore the reported enrichment / coefficient p-values (Fig 1B/2A/2B, Tables S1-S3) cannot be regenerated 1:1 without a DAC data-access agreement. Per the brief I did NOT substitute a surrogate public MS GWAS (e.g. FinnGen) - that would change the cohort and yield only a qualitative concordance, never the reported values, and would dress up a clean restricted-data drop. NOT attempted: building the GSE74912 annotations / LD scores in isolation (ungradeable without the GWAS) and the secondary integrative steps (GSE118189 subsets, DICE eQTL, PCHi-C, ENCODE/Roadmap ChIP-seq). All seven reported claims are captured in claims.tsv so a holder of the IMSGC data can complete the 1:1 check using scope.md as the recipe.

💻 Code ↗ 🗄 Data: GSE74912

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
    assessed: 2026-06-15 ⛓ bab749a77a23
✎ 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

Although MS GWAS variants are broadly enriched across peripheral immune cell types, it is unclear which specific cell populations independently and causally mediate these genetic effects; the paper tests whether the enrichment reflects shared regulatory landscapes or truly independent cell type-specific contributions to disease risk.

Core claims
  • CD4 T cells and B cells independently mediate MS GWAS genetic signals through their open chromatin regions, beyond shared regulatory landscapes. finding
  • Among CD4 T cell subsets, the Th17 subset independently drives the MS GWAS enrichment. finding
  • Within the B cell lineage, memory B cells drive the MS GWAS enrichment signal. finding
  • Immunomodulatory treatments (natalizumab, interferon, glatiramer acetate) attenuate/suppress chromatin accessibility signals at MS GWAS in driver cell types. finding
  • Stratified LD score regression (LDSC) with joint and pairwise models can dissect independent versus shared cell type-specific GWAS heritability enrichment in open chromatin regions. method
  • CD8 T cell and NK cell MS enrichments are largely explained by shared regulatory landscapes also present in CD4 T cells. finding
  • Findings replicate in immune cells sorted from untreated MS patients, with effector memory CD4 T cells and class-switched memory B cells showing independent enrichment. finding
  • Integration of statistical fine-mapping with chromatin interaction data (promoter capture HiC) nominates putative causal genes at MS loci. resource
Experimental setups
Assay System Perturbation Readout Platform
bulk ATAC-seq (open chromatin regions) integrated with MS GWAS via stratified LD score regression (LDSC) 16 flow-sorted hematopoietic progenitor and terminal cell populations from human peripheral blood or bone marrow none MS GWAS heritability enrichment in OCRs per cell type
bulk ATAC-seq + LDSC (joint and pairwise stratified models) human CD4 T cell subsets (naive effector, Th1, Th2, Th17, follicular Th, naive Tregs, memory Tregs) none LDSC coefficient (τc) p-value for independent heritability contribution
bulk ATAC-seq + LDSC (joint model) human B cell lineage subsets (naive B cells, memory B cells, plasmablasts) none LDSC coefficient p-value for independent heritability contribution
bulk ATAC-seq + LDSC flow-sorted CD4 T and B cell subsets from 6 untreated MS patients none (untreated MS) MS GWAS heritability enrichment and joint-model coefficient p-values
bulk ATAC-seq + LDSC (joint model treated vs untreated) CD4 T and B cell subsets from 3 MS patients under treatment drug (natalizumab, interferon, or glatiramer acetate) MS GWAS heritability enrichment attenuation in OCRs
LDSC heritability enrichment of comparator GWAS in hematopoietic OCRs GWAS of AD, SCZ, BPD, T1D, Crohn's, UC, SLE, RA, PBC across 16 hematopoietic OCR sets none cell type-specific heritability enrichment p-values
Key results
  • Strongest single-cell MS GWAS enrichment in CD4 T cell OCRs p = 1.47×10^-18
  • Strong MS GWAS enrichment in B cell OCRs p = 3.27×10^-15
  • Joint LDSC model shows B cells and CD4 T cells independently contribute to SNP heritability B: p = 3.99×10^-5; CD4 T: p = 3.49×10^-4
  • CD8 T cell OCRs no longer significant after stratifying on CD4 T cells coefficient p = 0.21
  • NK cell OCRs no longer significant after conditioning on CD4 T or CD8 T cells p = 0.165 (CD4); p = 0.356 (CD8)
  • Th17 cells independently contribute to MS heritability in CD4 joint model coefficient p = 4.69×10^-4
  • Memory B cells independently contribute to MS heritability in B cell joint model coefficient p = 1.10×10^-3
  • In untreated MS patients, effector memory CD4 T cells (T4em) independently enriched; treatment attenuates cMBc and T4em enrichments T4em joint coefficient p = 7.45×10^-3
Key statistics
  • pvalue 1.47×10^-18 (CD4 T cell OCR MS GWAS enrichment)
  • pvalue 4.00×10^-18 (CD8 T cell OCR MS GWAS enrichment)
  • pvalue 3.27×10^-15 (B cell OCR MS GWAS enrichment)
  • pvalue 4.23×10^-14 (NK cell OCR MS GWAS enrichment)
  • pvalue 4.17×10^-9 (monocyte OCR MS GWAS enrichment (ameliorated by conditioning on CD4 T or B cells))
  • count 41,505 MS cases vs 38,242 RA cases (comparable GWAS sample sizes, MS enrichments stronger)
  • pvalue 3.27×10^-4 (B cell-specific OCR MS GWAS enrichment (significant))
  • pvalue 2.58×10^-4 (class-switched classical memory B cell (cMBc) enrichment in untreated MS patients)

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.

The study applied stratified LD Score Regression (S-LDSC) to quantify enrichment of MS GWAS heritability in open chromatin regions (OCRs) from 16 flow-sorted hematopoietic cell populations drawn from published ATAC-seq datasets. Independent cell-type contributions were identified using a joint S-LDSC model (all annotations simultaneously) and 240 pairwise conditioned models, with the coefficient τ_c p-value as the primary test statistic. Analyses were replicated in ATAC-seq data from MS patient–derived cells (n=6 untreated, n=3 treated) and extended to nine comparator GWAS (autoimmune and neuropsychiatric diseases). Results were reported as exact p-values and visualised as −log10(p-value) throughout; confidence intervals on enrichment estimates were not reported in the text.

Replicationmixed Sample sizeMS GWAS: ~41,505 cases (from referenced published GWAS). Reference hematopoietic ATAC-seq: 16 flow-sorted populations from published datasets (refs 18–20); individual donor n not stated in main text. MS patient ATAC-seq: n=6 untreated, n=3 treated (clinical details in Table S13). No formal power calculation described. Groups16 hematopoietic cell types vs. MS GWAS heritability; CD4 T cell subsets vs. MS heritability; B cell lineage subsets vs. MS heritability; untreated vs. treated MS patients; MS vs. nine comparator diseases Pairingna Randomization/blindingnot stated Dispersionnone Exact p-valuesyes Effect sizesyes Confidence intervalsno Multiplicity correctionBonferroni correction
Statistical tests used
Test Applied to n Assumptions
Stratified LD Score Regression (S-LDSC), marginal per-cell-type enrichment Heritability enrichment of MS GWAS in OCRs of 16 hematopoietic cell populations (Fig. 1B; Table S1) MS GWAS summary statistics (~41,505 cases per referenced GWAS); donor n for ATAC-seq datasets not stated in main text not stated
S-LDSC joint model — coefficient τ_c p-value (all 16 OCR annotations + baseline simultaneously) Independent per-cell contribution to MS heritability controlling for all other cell types (Fig. 2A; Table S2) Same GWAS summary statistics; 16 cell-type annotations not stated
S-LDSC pairwise conditioned model — coefficient τ_c p-value Each of 16 cell types stratified against each of the other 15 individually (Fig. 2B; Table S3; 240 total comparisons) Same GWAS summary statistics not stated
S-LDSC on cell type-specific OCRs only (peaks unique to each terminal cell type) Whether exclusively cell-specific peaks enrich for MS heritability (Fig. S2; Table S4) Not stated not stated
S-LDSC marginal enrichment and joint model on CD4 T cell subsets (Th1, Th2, Th17, follicular Th, naïve/memory Tregs) CD4 T cell subset analyses to identify Th17 as driver (Fig. 3B–C; Tables S7–S8) Not stated not stated
S-LDSC marginal enrichment and joint model on B cell lineage subsets (naïve, memory, plasmablasts) B cell subset analyses to identify memory B cells as driver (Fig. 4B–C; Tables S10–S11) Not stated not stated
S-LDSC marginal enrichment and joint model on MS patient–derived CD4 T and B cell subsets Replication in untreated (n=6) and treated (n=3) MS patients (Fig. 5; Tables S14–S19) n=6 untreated MS patients; n=3 treated MS patients (glatiramer acetate, interferon, or natalizumab) not stated
S-LDSC marginal enrichment and joint model applied to nine comparator GWAS Cross-disease comparison (AD, SCZ, BPD, T1D, CD, UC, SLE, RA, PBC) in same 16 hematopoietic OCR annotations (Fig. S3; Tables S5–S6) Disease-specific published GWAS summary statistics; case n varies by disease not stated
Approaches that could also have been used
  • Bonferroni correction was applied separately to each family of S-LDSC tests (16-cell marginal, 240-pairwise, and cell-specific analyses).
    Could also: Benjamini-Hochberg false discovery rate (FDR) correction could also be applied within each test family. — Bonferroni controls the family-wise error rate and is most conservative; BH-FDR controls the expected proportion of false positives and typically yields greater power when many tests are correlated — relevant here because OCR annotations across immune cell types are substantially correlated, which can make Bonferroni overly stringent.
  • Independent cell-type contributions were assessed by entering all 16 OCR annotations simultaneously into a single joint S-LDSC model.
    Could also: A forward stepwise model-selection approach (adding one annotation at a time while it significantly improves fit) could also identify a minimal independent set of driving annotations. — With 16 correlated annotations in one regression, coefficient estimates can become unstable; a stepwise approach would yield a more parsimonious model and could more clearly separate annotations whose signal is fully absorbed by others from those that retain independent contribution.
  • S-LDSC enrichment estimates and τ_c coefficients were reported primarily as p-values and visualised without confidence intervals.
    Could also: Reporting LDSC enrichment fold-enrichment estimates and τ_c coefficients with standard errors or 95% CIs alongside p-values is also standard practice in the S-LDSC literature. — CIs on effect estimates convey both magnitude and precision of enrichment, enabling direct comparison of effect sizes across cell types and diseases beyond what p-values alone communicate.
  • Bulk ATAC-seq from flow-sorted populations was used to define OCR annotations for S-LDSC.
    Could also: Single-cell ATAC-seq (scATAC-seq) from the same or comparable populations could also define cell-type OCR annotations. — Bulk ATAC-seq reflects the average chromatin landscape of a sorted population; scATAC-seq would additionally resolve within-population heterogeneity and could surface rare subpopulations whose open chromatin signal is diluted in bulk profiles.
  • Heritability enrichment in functional annotations was quantified using S-LDSC.
    Could also: Alternative enrichment frameworks such as GARFIELD, GoShifter, or GREGOR could also test for enrichment of GWAS signals in the same OCR annotations. — These methods use permutation-based or linkage-based frameworks rather than LD-score regression and make different modelling assumptions; applying one as a sensitivity analysis would constitute independent technical replication of the enrichment findings.
  • The patient-derived replication used n=6 untreated and n=3 treated MS patients with no stated power calculation.
    Could also: A larger, prospectively designed matched cohort (e.g., pre-treatment / post-treatment samples from the same individuals) with a pre-specified power calculation could also be used. — With n=3 treated patients drawn from three different therapies, the treatment-effect analyses have limited statistical power and cannot distinguish drug-specific from shared effects; a matched longitudinal design would strengthen causal inference about treatment-associated chromatin changes.
Software: S-LDSC (stratified LD Score Regression) · ATAC-seq processing pipeline (described in refs 18 and 21) · Statistical fine-mapping (referenced in methods overview; specific tool not named in available text) · Promoter capture Hi-C (referenced for chromatin interaction integration; specific software not named in available text)

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
19
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.

GSM997225 GEO in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
rs1465697 RefSNP in Results (http://purl.org/orb/Results)
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.md — pmid-35672799

Paper: Guo MH et al. "Dissection of multiple sclerosis genetics identifies B and CD4+ T cells as driver cell subsets." Genome Biol 2022;23:139. PMCID PMC9175345 · DOI 10.1186/s13059-022-02694-y.

Declared code: https://github.com/bulik/LDSC (the stratified LD-score regression tool — a third-party method, not the authors' own pipeline; P16-valid). Declared data: GEO GSE74912 (Corces et al. hematopoiesis ATAC-seq).

The pipeline (as described in Methods)

The headline computational result is cell-type-specific stratified LD-score regression (S-LDSC) of MS GWAS heritability over open-chromatin-region (OCR) annotations:

  1. Take ATAC-seq for 16 human hematopoietic populations (GSE74912, Corces 2016).
  2. Call peaks; for each cell type derive cell-type-specific OCRs (peaks not overlapping any other cell type's peaks).
  3. Build LDSC annotations from those OCRs against the baseline model (original 53-annotation model), reference = 1000 Genomes Phase 1 European, HapMap3 SNPs, regression weights from the LDSC data release.
  4. Run S-LDSC of the MS GWAS summary statistics (IMSGC 2019 discovery set; 8,278,136 variants; 14,802 cases / 26,703 controls) in three modes:
    • single-annotation enrichment → Fig 1B / Table S1
    • joint model over all 16 cell types (coefficient τ_c) → Fig 2A / Table S2
    • pairwise conditioning (one cell type on another) → Fig 2B / Table S3

In scope (pipeline-derived, would be gradeable)

  • The S-LDSC enrichment p-values per cell type (Fig 1B / Table S1).
  • The joint-model coefficient p-values (Fig 2A / Table S2).
  • The pairwise conditional coefficient p-values (Fig 2B / Table S3). All three require both the GSE74912-derived annotations and the MS GWAS summary statistics as the dependent variable.

Out of scope

  • Wet-lab ATAC-seq of MS patient immune cells (generated in-house, not a public-pipeline reproduction).
  • Downstream eQTL (DICE), PCHi-C, ChIP-seq (ENCODE/Roadmap) integrative steps — secondary, not the headline claim.
  • CD4/B subset analyses using GSE118189 (secondary, same GWAS dependency).

Feasibility verdict — BLOCKED by restricted data

dependency public? evidence
LDSC code (bulik/ldsc) yes GitHub public
GSE74912 ATAC-seq yes GEO resolves, 130 samples, public supplementary peaks
LDSC baseline + 1000G ref + weights yes Broad/Alkes-group public data release
MS GWAS summary statistics (GCST009597) NO — request-only GWAS Catalog study GCST009597 fullPvalueSet=false (no hosted sumstats); imsgc.net/downloads = request-only form + Data Access Committee approval, "no redistribution"; the paper itself states "MS GWAS summary statistics are available via request to the IMSGC"

Every gradeable claim (Fig 1B, 2A, 2B / Tables S1–S3) is a function of the MS GWAS as the dependent variable. The ATAC half is public, but on its own the paper reports no standalone quantitative number (e.g. per-cell-type peak count) that could be reproduced and graded 1:1.

Outcome: drop, drop_reason = data_restricted. The essential input (MS GWAS sumstats, the paper's own data) is controlled-access / on-request, so the reported heritability-enrichment values cannot be regenerated without a Data Access Committee agreement. No fabricated/surrogate result is substituted. A surrogate public MS GWAS (e.g. FinnGen) would change the cohort and could only give a qualitative concordance check, never the reported p-values — per the brief ("honest 1:1; do not chase the last 20%") this was deliberately not done.

Figures / tables: Fig 1B1 TableFig 2AFig 2B
C1
Reported
CD4 T cell OCR enrichment p = 1.47e-18 (Fig 1B / Table S1)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C2
Reported
CD8 T cell OCR enrichment p = 4.00e-18 (Fig 1B / Table S1)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C3
Reported
B cell OCR enrichment p = 3.27e-15 (Fig 1B / Table S1)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C4
Reported
B cell joint-model coefficient p = 3.99e-5 (Fig 2A / Table S2)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C5
Reported
CD4 T cell joint-model coefficient p = 3.49e-4 (Fig 2A / Table S2)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C6
Reported
B cell pairwise coefficient vs HSC p = 1.47e-12 (Fig 2B / Table S3)
Reproduced
not reproduced - MS GWAS access-restricted
partial
C7
Reported
B cell pairwise coefficient vs CD4 T p = 8.33e-5 (Fig 2B / Table S3)
Reproduced
not reproduced - MS GWAS access-restricted
partial

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 44/100

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: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
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 🟡
Input / endpoint not comparable 1:1
+1 pts
From: Q1 · Data identity 🔴
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +7

This is a clean data_restricted drop, not a discrepancy. Every one of the seven reported S-LDSC enrichment/coefficient p-values (C1–C7) is computed against the IMSGC-2019 MS GWAS summary statistics (GCST009597), which are controlled-access (GWAS Catalog fullPvalueSet=false; imsgc.net request-only behind a Data Access Committee). The method (bulik/ldsc) and the chromatin half of the input (GSE74912) are public but produce no standalone reported number, so 0/7 claims are gradeable. The block sits on data availability / our access, not on the authors (they document the request route correctly) and not on a numeric or fabrication problem — hence q5/q7 yellow rather than red, with q1/q2 red capturing the non-availability.

🤝
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.

80.2 k
tokens (I/O) · 4.5 M incl. cache
8 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.