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

Structural modeling and functional characterization of a novel gain-of-function TLR8 variant causing severe inflammatory syndrome

· 2026
PubMed 41729082 ↗ pmid-41729082
L1 89/100 3/4
Why this verdict

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

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Same input data as the authors
  • 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
What did not (or only partly)
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
89/100
Reproducibility score
0.8 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 77% of all assessed papers rank 246 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

INTERIM (MD running). Paper's pipeline-derived computational core reproduces 1:1. All 6 in-silico variant predictors match exactly (gnomAD absent; PolyPhen-2 HumDiv 0.043 / HumVar 0.004 reproduced to 3 sig-figs incl sens/spec on the live server, confirming UniProt Q9NR97; SIFT tolerated; MutationTaster benign; CADD deleterious-leaning consistent with 'proxy-deleterious'). The HEADLINE computational claim - DynaMut normal-mode ddG - reproduces EXACTLY for both states (inactive 3W3G -0.238 destabilizing, active 4R08 +0.047 stabilizing). Static interface crystal-water analysis qualitatively reproduces Fig 6B's active-specific extra water-bridge. Tier-4 OpenMM MD (Fig 6D) is now genuinely RUNNING on «our HPC» H100 («job»). NOT attempted: wet-lab assays (out of scope) and RNA-seq DE (in-scope but blocked - data on-request).

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 89
    assessed: 2026-06-19 ⛓ 3589cd32b2c8
✎ 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
2026-06-26
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 study tests whether a previously unreported hemizygous TLR8 missense variant (c.1552G>A, p.A518T), found in two male siblings with severe recurrent infections and systemic inflammation, is a pathogenic gain-of-function mutation, using an integrative approach combining clinical, genetic, structural, and functional data.

Core claims
  • TLR8 A518T is a gain-of-function variant that enhances NF-κB activation and increases secretion of proinflammatory cytokines upon stimulation compared with WT TLR8 finding
  • The A518T mutant TLR8 protein shows reduced baseline abundance due to faster turnover and increased proteasomal degradation finding
  • Computational modeling predicts that A518T stabilizes the active TLR8 homodimer interface via additional water-mediated hydrogen bonds near the uridine-binding site mechanism
  • Siblings carrying TLR8 A518T display a systemic proinflammatory immune signature, including elevated cytokines, chemokines, interferons, and ISG expression finding
  • Standard in silico prediction tools (PolyPhen-2, MutationTaster, SIFT) classified the variant as benign or tolerated, failing to capture its gain-of-function effect, unlike CADD which flagged it as proxy-deleterious finding
  • An integrative approach combining clinical, genetic, structural, and functional data can establish pathogenicity of novel inborn-errors-of-immunity variants method
  • TLR8 A518T carriers exhibit dysregulated B cell maturation/differentiation and T cell exhaustion, resulting in defective humoral immunity finding
  • X chromosome inactivation in the heterozygous mother is random rather than skewed finding
Experimental setups
Assay System Perturbation Readout Platform
targeted exome sequencing P1, P2, and mother (patient blood/DNA) none identification of shared hemizygous/heterozygous TLR8 c.1552G>A variant
plasma cytokine/chemokine/interferon profiling plasma from P1 and P2 none levels of IL-6, TNF, GM-CSF, CXCL10, CCL2, CCL19, CCL20, IFN-α, IFN-λ, IFN-β, IFN-γ
quantitative PCR for interferon-stimulated genes and IRF7 PBMCs from P1, P2, mother vs. age-matched controls none mRNA expression levels of ISGs, IRF7, and TLR8
multiparametric flow cytometry immunophenotyping PBMCs from P1, P2, mother vs. age-matched controls none frequencies of T cell, B cell, monocyte, and dendritic cell subsets
in vitro B cell differentiation assay B cells isolated from P1 and P2 CD40L stimulation differentiation into plasmablasts/memory cells and IgG/IgM production
NF-κB activation and cytokine secretion assay cells expressing WT or A518T TLR8 TLR8 A518T variant expression plus ligand stimulation NF-κB activation and proinflammatory cytokine secretion
protein degradation/turnover assay cells expressing WT or A518T TLR8 TLR8 A518T variant expression protein abundance, turnover rate, proteasomal degradation
computational structural modeling (homology modeling, normal mode analysis, molecular dynamics) TLR8 homodimer protein structure (in silico) A518T substitution structural stability (ΔΔG), hydrogen bonding, vibrational entropy MODELLER v10.3, DynaMut
Key results
  • TLR8 A518T enhanced NF-κB activation and increased secretion of proinflammatory cytokines compared with WT TLR8 upon stimulation
  • Mutant TLR8 protein showed reduced abundance due to faster turnover and increased proteasomal degradation
  • A518T destabilized the inactive TLR8 conformation but stabilized the active homodimer conformation ΔΔG = -0.238 kcal/mol (inactive) vs +0.047 kcal/mol (active)
  • Elevated plasma levels of proinflammatory cytokines, growth factors, and chemokines in P1 and P2, with elevated IFN-α/IFN-λ in P2 and elevated IFN-γ in both
  • Upregulated expression of interferon-stimulated genes and IRF7 in PBMCs of both siblings versus age-matched controls
  • PolyPhen-2, MutationTaster, and SIFT classified the variant as benign/tolerated despite its functional gain-of-function effect
  • Both siblings showed increased naive/transitional B cells and reduced memory B cells and plasmablasts, indicating impaired B cell maturation
  • Delayed in vitro B cell differentiation into plasmablasts/memory cells with reduced IgG/IgM production after CD40L stimulation
Key statistics
  • other ΔΔG = -0.238 kcal/mol (predicted destabilization of inactive TLR8 conformation by A518T)
  • other ΔΔG = +0.047 kcal/mol (predicted stabilization of active TLR8 homodimer conformation by A518T)
  • other PolyPhen-2 HumDiv score 0.043 (sensitivity 0.94, specificity 0.83) (benign classification of variant)
  • other PolyPhen-2 HumVar score 0.004 (sensitivity 0.98, specificity 0.35) (benign classification of variant)
  • other X chromosome inactivation ratios: AR = 0.53, RP2 = 0.49 (random X inactivation pattern in mother's PBMCs)
  • count <100 cells per μL (severe leukopenia in both siblings)
  • other P1 age 27, P2 age 24, mother age 48; plasma profiling performed at ages 25 (P1) and 22 (P2) (study participant ages)

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 is a case report describing two male siblings (and their heterozygous carrier mother) with a novel TLR8 A518T variant, using an integrative but largely descriptive approach: clinical/laboratory characterization, plasma cytokine and immune-cell profiling (flow cytometry, qPCR), several in silico variant-effect predictors (CADD, PolyPhen-2, MutationTaster, SIFT), and computational structural modeling/molecular dynamics (MODELLER, DynaMut) comparing wild-type and variant TLR8. Results are reported mainly as qualitative or point-estimate comparisons (e.g., 'significantly elevated,' ΔΔG values) against age-matched controls or wild-type protein, without named formal statistical tests, exact p-values, or dispersion measures appearing in the excerpted text.

Replicationbiological Sample sizeTwo affected male siblings (P1, P2) and their heterozygous mother (M) are the study subjects; comparisons are made to 'age-matched controls' or 'healthy controls' whose number is not specified in the provided text; no power calculation is described GroupsTLR8 A518T variant carriers (2 siblings, and to a lesser extent their mother) vs. age-matched/healthy controls; and WT vs. A518T TLR8 protein in structural/functional assays Pairingunclear Randomization/blindingnot stated Dispersionunclear
Statistical tests used
Test Applied to n Assumptions
not specified (differences described qualitatively, e.g. 'significantly elevated/reduced') plasma cytokine/chemokine levels, ISG expression by qPCR, and immune cell subset frequencies by flow cytometry in patients vs. age-matched controls 2 affected siblings (P1, P2) plus their mother; number of age-matched controls not stated in the text not stated
point-estimate free energy calculation (ΔΔG) via DynaMut normal mode analysis comparison of WT vs. A518T TLR8 stability in active and inactive homodimer conformations (Figure 6C) single computational model per conformation/genotype not stated
qPCR-based ratio calculation (not a formal statistical test) X chromosome inactivation pattern in mother's PBMCs using AR and RP2 loci single sample per locus (ratios 0.53 and 0.49) not stated
Approaches that could also have been used
  • Differences in cytokine/chemokine levels, ISG expression, and immune cell subset frequencies between the variant carriers and controls are described qualitatively (e.g., 'significantly elevated') without a named statistical test in the excerpted text.
    Could also: Given the very small patient sample (n=2), a nonparametric approach such as comparison against a defined reference range/interval from a larger control cohort, or explicit reporting of exact p-values and effect sizes, could also be used. — This would let readers gauge the magnitude and uncertainty of reported differences beyond directional descriptors, which is especially informative when the patient group is very small.
  • X chromosome inactivation in the mother was assessed using AR and RP2 qPCR ratios (0.53 and 0.49) interpreted directly as indicating a random pattern.
    Could also: Reporting these ratios alongside an established statistical threshold or confidence interval for skewing (as commonly used in Lyon-type XCI assays) could also be used. — This would make explicit the statistical basis for the 'random vs. skewed' conclusion rather than relying on the ratio values alone.
  • Pathogenicity of the TLR8 variant was assessed with several in silico predictors (CADD, PolyPhen-2, MutationTaster, SIFT), which gave discordant classifications (benign/tolerated vs. proxy-deleterious).
    Could also: A formal variant-classification framework, such as ACMG/AMP quantitative scoring or additional meta-predictors (e.g., REVEL, AlphaMissense), could also be used to reconcile discordant computational predictions. — This provides a standardized, weighted approach for combining multiple lines of computational evidence when individual tools disagree.
  • Structural stability differences between WT and A518T TLR8 were summarized as single ΔΔG point estimates from DynaMut for the active and inactive conformations.
    Could also: Reporting ΔΔG with an associated confidence interval, or performing replicate molecular dynamics/DynaMut runs to obtain a mean ± SD across replicates, could also be used. — This would convey the uncertainty inherent in single-model structural stability predictions.
  • Multiple immune parameters (cytokines, ISGs, cell subset frequencies) were evaluated across the same small set of individuals without a described multiplicity correction.
    Could also: Applying a false discovery rate procedure (e.g., Benjamini-Hochberg) across the panel of measured markers could also be used. — This would help indicate which of the many parallel comparisons are most robust when several markers are tested in the same individuals.
  • Quantitative findings (e.g., cytokine levels, qPCR fold changes) are presented without a stated measure of dispersion (SD, SEM, CI, or IQR) in the excerpted text.
    Could also: Reporting values with an explicit spread measure, such as an interquartile range or 95% confidence interval appropriate for small n, could also be used. — This would allow readers to assess the precision of point estimates rather than relying on categorical descriptors like 'elevated' or 'reduced.'
Software: MODELLER v10.3 · DynaMut · CADD (Combined Annotation Dependent Depletion) GRCh38-v1.4 · PolyPhen-2 · MutationTaster · SIFT

What was reproduced

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

Scope — pmid-41729082

Title: Structural modeling and functional characterization of a novel gain-of-function TLR8 variant causing severe inflammatory syndrome Journal: JCI Insight 2026;11(4):e187422 · DOI 10.1172/jci.insight.187422 · PMCID PMC12956005 · CC BY (open access) Lab: Altfeld lab, Leibniz Inst. of Virology / UKE Hamburg.

Nature of the paper

Single-family case study (2 male siblings P1/P2 + mother M) of a novel hemizygous TLR8 missense variant c.1552G>A, p.A518T (chrX:12920592 GRCh38; NM_138636.5; NP_619542.1; rs2147259148; ClinVar RCV002263534.1). It integrates clinical, genetic, computational structural modeling, and in-vitro functional data. The bulk of the paper is wet-lab. Only the computational/in-silico parts are pipeline-derived and in scope.

IN SCOPE (pipeline-derived computational results)

id result tool(s) reported value paper loc
P-CADD variant pathogenicity CADD GRCh38-v1.4 "proxy-deleterious" (exact score in Suppl. Table 1) Results "Identification…"
P-PP2div PolyPhen-2 HumDiv PolyPhen-2 score 0.043; sens 0.94; spec 0.83 → benign Results "Identification…"
P-PP2var PolyPhen-2 HumVar PolyPhen-2 score 0.004; sens 0.98; spec 0.35 → benign Results "Identification…"
P-SIFT SIFT SIFT "tolerated" Results "Identification…"
P-MT MutationTaster MutationTaster "benign" Results "Identification…"
P-gnomad population frequency gnomAD "not present in gnomAD" Results "Identification…"
P-ddg-in NMA stability, inactive DynaMut (on 3W3G) ΔΔG = -0.238 kcal/mol (destabilizing) Results "Computational modeling…"
P-ddg-ac NMA stability, active DynaMut (on 4R08) ΔΔG = +0.047 kcal/mol (stabilizing) Results "Computational modeling…"
P-hbond extra water-mediated H-bonds at dimer interface (active) MODELLER v10.3 + ChimeraX v1.7 qualitative: A518T forms additional water-mediated H-bonds at interface w/ uridine (Fig 6B) Results / Fig 6B
P-md water-mediated interchain H-bonds over 100ns OpenMM 8.1.2 MD (4 systems) qualitative trend: A518T > WT in active state; A518T < WT in inactive state, whole trajectory (Fig 6D) Results / Fig 6D

Structural inputs (datasets, see dataset_profile.json):

  • PDB 4R08 — active/liganded TLR8 ectodomain homodimer (+uridine agonist).
  • PDB 3W3G — inactive/unliganded TLR8 ectodomain (with Z-loop).

OUT OF SCOPE (wet-lab / manual / not pipeline-derived)

  • Targeted exome sequencing (variant calling) — raw data not deposited; variant itself is the only output, already public in dbSNP/ClinVar.
  • Flow-cytometry immunophenotyping, plasma cytokine profiling, qPCR ISGs, X-inactivation assay (Figs 1–5, 7A/B; Tables 1–3).
  • NF-κB / cytokine functional assays, Western blots, cycloheximide chase, ubiquitination / proteasome assays (Figs 8–9).

PARTIALLY in scope but BLOCKED (data not available)

  • RNA-seq DE (Fig 7C/D, Suppl Fig 10): "515 vs 98 genes" (unstim) and "55 vs 9 genes" (stim) differentially expressed. This is pipeline-derived (DESeq2-style DE + pathway enrichment) but no sequencing-data accession is given — "All data are available in the main text or the supplemental materials… corresponding author will respond to emailed requests." → raw counts/FASTQ unavailable → drop_reason for this sub-result: data_restricted (on-request). Not attempted.

Reproduction plan (priority)

  1. Tier 1 (floor, exact numbers): in-silico predictors — gnomAD, CADD, PolyPhen-2 (HumDiv/HumVar), SIFT, MutationTaster. Deterministic, exact compare.
  2. Tier 2 (floor, exact numbers): DynaMut ΔΔG on 4R08 + 3W3G with A518T.
  3. Tier 3 (harder): MODELLER homology model of A518T on 4R08/3W3G + count water-mediated interface H-bonds (3.5 Å / 150°) WT vs variant.
  4. Tier 4 (heavy, keep-going): OpenMM 100 ns MD ×4 systems on «our HPC» GPU, water-mediated intercha
Figures / tables: TableFig 6BFig 6D
C1_gnomad
Reported
not present in gnomAD
Reproduced
Variant not found in gnomAD v4 (exome+genome) for X-12920592-G-A (GraphQL null)
exact
C2_cadd
Reported
proxy-deleterious by CADD (GRCh38-v1.4)
Reproduced
CADD GRCh38-v1.4 PHRED 19.83 / raw 2.0185 (deleterious-leaning)
within tolerance
C3_pp2_humdiv
Reported
benign; 0.043 / sens 0.94 / spec 0.83
Reproduced
PolyPhen-2 HumDiv live v2.2.3r408: 0.043 / sens 0.943 / spec 0.829 (benign)
exact
C4_pp2_humvar
Reported
benign; 0.004 / sens 0.98 / spec 0.35
Reproduced
PolyPhen-2 HumVar live v2.2.3r408: 0.004 / sens 0.977 / spec 0.351 (benign)
exact
C5_sift
Reported
tolerated
Reproduced
tolerated (dbNSFP SIFT pred T 0.12-0.13; VEP 0.05)
exact
C6_mutationtaster
Reported
benign
Reproduced
polymorphism (pred N, score 1.0) = benign (dbNSFP/MutationTaster)
exact
C7_ddg_inactive
Reported
ddG = -0.238 kcal/mol (destabilizing)
Reproduced
DynaMut headline ddG = -0.238 kcal/mol (Destabilizing) [3W3G A518T chain A]
exact
C8_ddg_active
Reported
ddG = +0.047 kcal/mol (stabilizing)
Reproduced
DynaMut headline ddG = +0.047 kcal/mol (Stabilizing) [4R08 A518T chain A]
exact
C9_hbond_static
Reported
A518T forms additional water-mediated H-bonds at interface, stabilizing active homodimer (Fig 6B)
Reproduced
Static crystal-water interchain bridges (3.5A): active +1 NEW via Thr518-OG1, inactive 0 -> same active-specific asymmetry
partial
C10_md_trend
Reported
water-mediated interchain H-bonds increased A518T>WT (active), decreased A518T<WT (inactive) over 100ns (Fig 6D)
Reproduced
RUNNING: OpenMM 8.1.2 MD on «our HPC» H100, 4 systems WT/A518T x active/inactive, 20ns each («job»); interchain water-bridge trajectory trend pending
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 89/100

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.

727.3 k
tokens (I/O) · 90.2 M incl. cache
164 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.