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

Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets.

Nat Immunol · 2023
L1 93/100 PQI 98
Why this verdict

The main results reproduced: recomputed values matched the published ones within tolerance.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Reported values were directly comparable
  • 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)
  • 🟡Could not use the authors’ exact input data
How its reproducibility compares
93/100
Reproducibility score
1.1 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 85% of all assessed papers rank 154 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 -> 1:1 REPRODUCED. The brief's enriched code/data were wrong (kauralasoo/eQTL-Catalogue-resources + GSE16879 belong to neither this paper); the real artifacts are github.com/jinghuazhao/INF (SCALLOP-INF pipeline) and GWAS Catalog full summary statistics GCST90274758-GCST90274848 (public, GRCh37, N~14736). Two reproductions: (A) the shipped sentinel table INF1.merge.cis.vs.trans reproduces the abstract numbers EXACTLY -- 180 pQTLs, 59 cis, 121 trans, 70 proteins, 59 with a cis-pQTL; (B) for 6 showcase cis-pQTLs (OPG, CXCL11, TNFB/LTA, CD40, CXCL5, TRAIL) we independently re-derived -log10 p from the deposited beta/SE at each reported sentinel -- all 6 exist at the exact GRCh37 position, all are genome-wide significant (p<5e-10), and the recomputed strengths match the reported values to <0.5% on the log scale; 5/6 sentinels are the marginal-min lead in the +/-1Mb cis window (CXCL5's lead rs425535 is ~6kb from the reported rs450373, same locus, essentially tied). No fabrication concern: the reported pQTL signals are fully derivable from the public deposited data. NOT ATTEMPTED (hard last ~20% / restricted): full pQTL discovery meta-analysis from raw cohorts (individual-level biobank data restricted), conditional/COJO 227-signal count (needs LD ref + GCTA), and colocalization/Mendelian-randomization (Figs 3-6, multi-stage).

💻 Code ↗ 🗄 Data: GSE16879

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 93
    assessed: 2026-06-15 ⛓ bf757fa4650e
✎ 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

Identifying the genetic determinants (pQTLs) of inflammation-related circulating plasma proteins will yield insights into physiology and the etiology of immune-mediated diseases and help prioritize therapeutic drug targets.

Core claims
  • Genome-wide pQTL mapping of 91 inflammation-related plasma proteins in 14,824 participants identified 180 pQTLs (59 cis, 121 trans). finding
  • This pQTL resource enables drug target prioritization by linking proteins between genotype and disease phenotype. resource
  • Mendelian randomization and colocalization identify proteins causally involved in immune-mediated disease etiology, revealing shared and distinct effects across diseases. method
  • CD40 has directionally discordant effects on risk of rheumatoid arthritis versus multiple sclerosis and inflammatory bowel disease. finding
  • Lymphotoxin-α (LTA) is implicated in multiple sclerosis pathogenesis via integration of pQTL with disease GWAS. mechanism
  • CXCL5 is implicated in the etiology of ulcerative colitis, supported by elevated gut CXCL5 transcript expression in UC patients. finding
  • Genetic variation in NLRC4 alters its expression and inflammasome activity, affecting circulating IL-18 levels (trans-pQTL via caspase-1 cleavage of pro-IL-18). mechanism
  • At least 50% of cis-pQTLs may be driven by underlying cognate cis-eQTLs, identified via colocalization across tissues. finding
Experimental setups
Assay System Perturbation Readout Platform
Genome-wide pQTL mapping (GWAS of plasma protein levels) Plasma from 14,824 European-ancestry participants across 11 cohorts none Plasma abundance of 91 inflammation-related proteins; genetic associations Olink Target Inflammation panel
pQTL replication Plasma from 1,585 ARISTOTLE participants none Replication/concordance of pQTL effect estimates Olink
pQTL replication Plasma from 35,556 Icelanders (deCODE) none Replication of locus–protein associations SomaScan (aptamer-based)
cis-eQTL colocalization analysis Whole blood (eQTLGen, n=31,684) none Colocalization of eQTL with pQTL; gene expression vs protein
Multi-tissue cis-eQTL colocalization analysis Multiple tissues/cell types (GTEx v.8, eQTL Catalogue) none Colocalization of cis-eQTLs with cis-pQTLs across tissues
trans-pQTL mediator identification (bioinformatic) pQTL meta-analysis data plus annotation databases none Prioritized candidate mediating genes ProGeM; STRINGdb
Mendelian randomization and colocalization pQTL data integrated with disease GWAS none (genetic instruments) Causal effects of proteins on immune-mediated disease risk
Gut transcript expression analysis Gut tissue from ulcerative colitis patients disease (UC vs control) CXCL5 transcript expression level
Key results
  • 180 significant pQTLs identified between 108 genomic regions and 70 proteins (59 cis, 121 trans) 180 pQTLs
  • Conditional analyses revealed 47 additional independent signals, raising total to 227 (99 cis, 128 trans) 227 signals
  • Strong correlation between pQTL effect estimates in ARISTOTLE and discovery meta-analysis r=0.97 (cis r=0.99, trans r=0.94)
  • 126 (71%) of 178 testable pQTLs replicated in ARISTOTLE or deCODE at P≤2.8×10^-4 71%
  • trans-pQTL hotspot at SH2B3 (rs3184504) associated with six proteins (CXCL9, CXCL10, CXCL11, CD5, CD244, IL-12B) 6 proteins
  • For 70 of 91 proteins ≥1 pQTL detected; 59 had a cis-pQTL 77% / 65%
  • Only 6 of 40 blood cis-eQTLs colocalized with cognate cis-pQTLs; IL18 eQTL-pQTL directionally discordant 6 of 40
  • Elevated gut CXCL5 transcript expression in patients with ulcerative colitis
Key statistics
  • correlation r = 0.97 (Pearson correlation of pQTL effect estimates ARISTOTLE vs discovery)
  • correlation cis r=0.99, trans r=0.94 (Replication effect-size correlation by pQTL type)
  • pvalue P ≤ 5 × 10^-10 (Genome-wide significance threshold for discovery pQTLs)
  • count 180 pQTLs (59 cis, 121 trans) (Total significant locus–protein associations)
  • count 227 signals (99 cis, 128 trans) (Total after conditional analysis (+47 independent))
  • other 0.003 (NTF3) to 0.285 (CCL8) (Proportion of variance explained by sentinel variants)
  • count 32 / 75 significant at P≤5×10^-10 (Replicated pQTLs in ARISTOTLE (32 of 174) and deCODE (75 of 158))
  • count 40 of 59 cis-pQTLs had a significant blood cis-eQTL; 32 colocalized in ≥1 tissue (cis-eQTL support for cis-pQTLs)

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 genome-wide protein quantitative trait locus (pQTL) study of 91 inflammation-related plasma proteins (Olink Target platform) in 14,824 European-ancestry participants across 11 cohorts. Genetic associations were estimated with linear regression per cohort and combined by fixed-effect meta-analysis, with significance set at a stringent genome-wide threshold (P ≤ 5 × 10^-10); downstream integration used statistical colocalization, eQTL comparison and Mendelian randomization to infer causality. Results were reported as effect estimates with two-sided P values, and findings were assessed for replication in independent cohorts (ARISTOTLE and deCODE).

Replicationunclear Sample sizeTotal discovery sample of 14,824 participants across 11 cohorts; replication cohorts ARISTOTLE (n = 1,585) and deCODE (n = 35,556); no explicit power calculation described in the provided text GroupsGenotype groups vs continuous plasma protein abundance (pQTL mapping); also disease GWAS integration Pairingna Randomization/blindingna Dispersionnone Exact p-valuesyes Effect sizesyes Multiplicity correctionStringent genome-wide significance threshold (P ≤ 5 × 10^-10) for discovery; Bonferroni-corrected threshold (P ≤ 2.8 × 10^-4) for replication
Statistical tests used
Test Applied to n Assumptions
Linear regression (per-cohort genetic association), combined by fixed-effect meta-analysis Genome-wide pQTL discovery across 91 proteins; significance P ≤ 5 × 10^-10 14,824 participants not stated
Pearson correlation Correlation of pQTL effect estimates between discovery meta-analysis and ARISTOTLE replication (r = 0.97 overall; cis r = 0.99, trans r = 0.94) 174 testable pQTL signals (ARISTOTLE n = 1,585) not stated
Linear regression (replication association testing) Replication in ARISTOTLE (n = 1,585) and deCODE (n = 35,556) at P ≤ 5 × 10^-10 and Bonferroni P ≤ 2.8 × 10^-4 ARISTOTLE 1,585; deCODE 35,556 not stated
Statistical colocalization (posterior probability ≥ 0.8) eQTL–pQTL pairs (eQTLGen, GTEx v8, eQTL Catalogue) for the 59 cis-pQTLs not stated
Mendelian randomization Assessing causal effects of proteins on immune-mediated disease risk (e.g., CD40, LTA, CXCL5) not stated
Conditional analysis Identifying additional independent pQTL signals (180 raised to 227) not stated
Approaches that could also have been used
  • Per-cohort estimates were combined using fixed-effect meta-analysis.
    Could also: A random-effects meta-analysis (or reporting between-cohort heterogeneity statistics such as I^2/Cochran's Q alongside the fixed-effect result) could also have been used. — Random-effects models additionally accommodate between-cohort heterogeneity arising from differing populations or assay conditions, and reporting heterogeneity helps readers gauge consistency of effects across the 11 cohorts.
  • Discovery significance used a fixed P-value threshold of 5 × 10^-10.
    Could also: A false discovery rate approach (e.g., Benjamini–Hochberg) could also have been applied across the protein–variant tests. — An FDR framework expresses the expected proportion of false discoveries among reported signals and can offer a complementary view of the discovery set's reliability alongside a fixed family-wise threshold.
  • Replication concordance was summarized with Pearson's correlation of effect estimates.
    Could also: Spearman's rank correlation or a Deming/weighted regression that accounts for measurement error in both axes could also have been used. — Rank-based correlation is robust to outliers and non-linearity, while errors-in-variables regression accounts for sampling uncertainty in both the discovery and replication estimates, which can refine the slope of agreement.
  • Causal inference for disease etiology relied on Mendelian randomization with colocalization support.
    Could also: Sensitivity analyses such as MR-Egger, weighted median, or formal pleiotropy/heterogeneity diagnostics could also accompany the primary MR estimates. — These complementary estimators relax different assumptions about horizontal pleiotropy and help characterize the robustness of a causal estimate to violations of the standard MR assumptions.
  • Colocalization used a posterior probability threshold of ≥ 0.8.
    Could also: Reporting the full posterior probability distribution or using methods that allow multiple causal variants (e.g., conditional/SuSiE-based colocalization) could also have been presented. — Showing the continuous posterior and allowing multiple causal signals can convey the strength of shared-variant evidence more granularly, especially in regions with allelic heterogeneity.

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
724
Impact: very 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.

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.

EFO_0004747 EFO in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
GSE16879 GEO in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
GSE206285 GEO in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
NCT04996797 NCT in Discussion (http://purl.org/orb/Discussion)
no other assessed paper uses this yet
NCT05013905 NCT in Discussion (http://purl.org/orb/Discussion)
no other assessed paper uses this yet
rs1800693 RefSNP in Discussion (http://purl.org/orb/Discussion)
no other assessed paper uses this yet
rs2364458 RefSNP in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
rs2364485 RefSNP in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet
rs5744292 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 — pmid-37563310

Paper: Zhao JH et al. (2023) Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets. Nat Immunol 24:1540–1551. PMID 37563310 · PMCID PMC10457199 · DOI 10.1038/s41590-023-01588-w. SCALLOP-INF consortium.

Metadata correction (brief enrichment was WRONG)

The scaffolded manifest.json lists code_url = kauralasoo/eQTL-Catalogue-resources and data_accession = GSE16879. Neither belongs to this paper (GSE16879 is the Arijs IBD/infliximab microarray set; the eQTL-Catalogue repo is Alasoo's resource). The auto-enrichment mismatched. Corrected pointers:

  • Code: https://github.com/jinghuazhao/INF (SCALLOP-INF analysis, branch master, active; the paper's own pipeline). cis/trans logic: cardio/cis.vs.trans.classification.R; sentinel detection: cardio/sentinels.R; shipped final pQTL table: rsid/10-4-2024/INF1.merge.cis.vs.trans.
  • Data: GWAS Catalog full summary statistics, public, one study per protein: GCST90274758–GCST90274848 (91 Olink Target-96 Inflammation proteins). GRCh37, GWAS-SSF v1.0, European, N≈14,736. FTP under .../summary_statistics/GCST90274001-GCST90275000/<GCST>/.

Study design (from abstract/results)

  • 91 plasma inflammatory proteins (Olink Target platform), 14,824 participants, meta-analysis across 11 cohorts (METAL).
  • Reported: 180 pQTLs (59 cis, 121 trans) across 70 proteins and 108 regions.
  • Genome-wide/study-wide significance threshold p < 5×10⁻¹⁰ (from repo: --clump-p1 5e-10, --cojo-p 5e-10, Manhattan abline(h=-log10(5e-10))).
  • cis = lead variant within ±1 Mb of the encoding gene (radius=1e6), else trans.

IN SCOPE (pipeline-derived, public data, low-hanging — attempted)

  1. Headline pQTL counts — reproduce 180 / 59 cis / 121 trans / 70 proteins from the shipped sentinel table INF1.merge.cis.vs.trans (the pipeline output) and re-classify cis/trans independently via the documented ±1 Mb rule.
  2. Independent sentinel verification vs deposited primary data — for a set of showcase cis-pQTLs (OPG, CXCL11, TNFB/LTA, CD40, CXCL5, TRAIL), download the protein's deposited GWAS-Catalog summary statistics and re-derive −log10 p from β/SE at the reported sentinel; confirm it is the genome-wide-significant lead in the cis window. This is the non-circular 1:1 check (deposited data → reported value).

OUT OF SCOPE (hard last ~20% / restricted / not attempted — why)

  • pQTL discovery meta-analysis from raw cohorts — individual-level genotype + Olink protein data are in restricted biobanks (INTERVAL, EstBB, KORA, …); not public → cannot rebuild the meta-analysis. (data_restricted for primary data; but the derived summary statistics ARE public, which is what we reproduce on.)
  • Exact conditional/COJO signal count (227 signals after conditioning) — needs an LD reference panel + GCTA-COJO per region; multi-step, beyond 80/20.
  • Colocalization (eQTLGen/GTEx) and Mendelian randomization / GSMR (Figs 3–6) — feasible in principle from public sumstats but multi-stage; not attempted here.
  • Wet-lab / manual / external-database curation results — out of scope by definition.

No completeness claim

We reproduce the headline pQTL counts and independently verify a sample of cis sentinels against the deposited primary summary statistics. We do NOT reproduce the full discovery meta-analysis, conditional signal counting, colocalization, or MR.

Figures / tables: table
C1
Reported
180 pQTLs
Reproduced
180
exact
C2
Reported
59 cis-pQTLs
Reproduced
59
exact
C3
Reported
121 trans-pQTLs
Reproduced
121
exact
C4
Reported
70 proteins with a pQTL
Reproduced
70
exact
C5
Reported
59 proteins with a cis-pQTL
Reproduced
59
exact
C6
Reported
OPG cis-pQTL log10p -49.47
Reproduced
-49.85 (rs2247769)
within tolerance
C7
Reported
CXCL11 cis-pQTL log10p -47.4
Reproduced
-47.62 (rs6827617)
within tolerance
C8
Reported
TNFB/LTA cis-pQTL log10p -613.38
Reproduced
-611.52 (rs2229092, p=3.0e-612)
within tolerance
C9
Reported
CD40 cis-pQTL log10p -281.61
Reproduced
-280.97 (rs1883832, p=1.1e-281)
within tolerance
C10
Reported
CXCL5 cis-pQTL log10p -171.05
Reproduced
-171.19 (rs450373, p=6.4e-172)
within tolerance
C11
Reported
TRAIL cis-pQTL log10p -47.4
Reproduced
-47.43 (rs574044675)
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 93/100

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.

163.5 k
tokens (I/O) · 12.7 M incl. cache
32 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.