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

Variants in LRRC7 lead to intellectual disability, autism, aggression and abnormal eating behaviors

· 2024
PubMed 39256359 ↗ pmid-39256359
L1 87/100 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.

Concordant (toward reproduced)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Every question reproduced
-1 pts
From: “every question reproduced”
Total score -5
✓ 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
  • Overall, the reproduction was clean
What did not (or only partly)
  • Every checked point held up.
How its reproducibility compares
87/100
Reproducibility score
0.7 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 72% of all assessed papers rank 301 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

PRELIMINARY (being finalized after C4 .msf grep completes). Clinical-genetics + wet-lab paper; most results out of scope (non-pipeline). Open-data computational claims reproduce: gnomAD pLI=1 + o/e=0.13 EXACT; all patient variants absent from gnomAD except p.Val1065Serfs*5 (exome AC=2) EXACT; CADD v1.7 within tolerance. BeviMed headline (100KGP) not attempted (data_restricted).

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 87
    assessed: 2026-06-18 ⛓ a40e0e4a0301
✎ 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
2026-06-22
Rubric version
v1.0
Assessed by
🤖 AI curator · v1.0 · run #1 2026-06-18
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 paper tests whether rare missense or loss-of-function variants in LRRC7 (encoding the postsynaptic scaffold protein Densin-180) are causally associated with a dominant neurodevelopmental disorder characterized by intellectual disability, autism, ADHD, aggression and abnormal eating behaviors.

Core claims
  • Heterozygous missense or loss-of-function variants in LRRC7 cause a dominant neurodevelopmental disorder in 33 identified individuals finding
  • High-impact rare variants in LRRC7 are statistically associated with intellectual disability in the 100,000 Genomes Project cohort finding
  • A PDZ domain variant (p.L1567Q) interferes with synaptic targeting of Densin-180 in primary cultured neurons finding
  • The LRR domain of Densin-180 binds new candidate interaction partners including protein phosphatase 1 (PP1) and MACF1 mechanism
  • Patient LRR missense variants reduce binding of Densin-180 to its LRR-domain interaction partners finding
  • Truncating variants in exons 20/21 are associated with hyperphagia-driven obesity in affected patients finding
  • C. elegans let-413 is functionally ancestral to the four human LAP genes and was used to model patient LRR variants method
  • Densin-180 is a postsynaptic scaffold linking cytoskeletal elements to signalling proteins such as αCaMKII mechanism
Experimental setups
Assay System Perturbation Readout Platform
Genetic association analysis (Bevimed) 100,000 Genomes Project cohort (77,539 participants) none posterior probability of association (PPA) with disease classes
Exome sequencing / panel-based sequencing GeneMatcher/confirmation cohort patients none identification of LRRC7 variants
Human phenotype ontology (HPO) phenotyping affected individuals across 9+ pedigrees none frequency of clinical phenotype terms
Yeast two-hybrid / BioID / co-immunoprecipitation 293T cells with tagged Densin-180 constructs LRR domain missense variants protein-protein interaction / binding to candidate partners (e.g. PP1, MACF1)
Immunofluorescence localization primary cultured neurons PDZ domain variant p.L1567Q synaptic targeting of Densin-180
CRISPR knock-in and phenotyping C. elegans (let-413 gene) L173M and L248P knock-in variants (homologous to human L221M/L296P) epithelial cell polarity / apical junction assembly phenotype
Structural modeling Densin-180 LRR region (in silico) none predicted 3D structure and position of variant residues AlphaFold server
Key results
  • LRRC7 ranked 19th of 42 genes associated with the Intellectual Disability disease class, with PPA nearly 1, second only to SHANK3 among synaptic genes PPA 1.0–9.3×10−6
  • 33 individuals (32 newly reported, 1 previously described) identified with LRRC7 variants and a neurodevelopmental disorder
  • High prevalence of ID, developmental delay, speech delay and atypical/aggressive behaviour among affected individuals ID 81%, NDD delay 84%, speech delay 81%, motor delay 45%, global developmental delay 30%, ASD 24%, ADHD 24%, atypical behaviour 90%
  • Aggression observed equally in male and female patients within the replication cohort 8 of 22 individuals
  • Four patients with truncating variants in exons 20/21 showed hyperphagia due to reduced satiety, three developing class II obesity 3 of 4 patients
  • Seizures and brain malformations observed in a subset of patients seizures in 6 patients, brain malformations in 5 patients
  • LRR missense variants disrupt conserved structural features (asparagine ladder, hydrophobic core) predicted to destabilise LRR folding
Key statistics
  • pvalue PPA 1.0–9.3×10−6 (LRRC7 association with Intellectual Disability disease class in 100KGP)
  • count 5529/29741 (probands assigned to the Intellectual Disability disease class in 100KGP)
  • count 33 individuals (total patients identified with LRRC7 variants)
  • other ID 81%, neurodevelopmental delay 84%, speech delay 81%, motor delay 45%, global developmental delay 30% (HPO-based phenotype frequencies in pLoF LRRC7 patients)
  • other ASD 24%, ADHD 24%, atypical/aggressive behaviour 90% (behavioural phenotype frequencies)
  • count 8 of 22 individuals (patients with aggression in replication cohort)
  • count 3 of 4 patients (patients with exon 20/21 truncating variants developing class II obesity)
  • count 6 patients with seizures; 5 with brain malformations (additional clinical features reported)

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 paper establishes statistical association between high-impact LRRC7 variants and intellectual disability using a Bayesian rare-variant gene-burden framework (BeviMed) applied to 29,741 probands in the 100,000 Genomes Project, reporting posterior probability of association (PPA) rather than frequentist p-values. This is complemented by a multicentric case series of 33 individuals assembled via GeneMatcher, with phenotypic characterisation reported as HPO-term frequencies (percentages). Functional evidence comes from C. elegans knock-in models and in vitro protein-interaction assays (yeast two-hybrid, BioID, co-immunoprecipitation); the statistical treatment of those experiments is not visible in the supplied text extract. Variant rarity in the general population was assessed descriptively against gnomAD allele frequencies rather than by formal enrichment testing.

Replicationmixed Sample sizeDiscovery: 100KGP with 29,741 probands (5,529 ID); Replication: 33 patients via GeneMatcher multicentric ascertainment. No formal power calculation stated. GroupsIndividuals with high-impact LRRC7 variants vs. all other probands in 100KGP (association phase); no formal control group in the case-series phase Pairingna Randomization/blindingnot stated Dispersionnone Exact p-valuesno Effect sizesno Confidence intervalsno Multiplicity correctionBeviMed's Bayesian framework inherently accounts for testing multiple genes across multiple disease classes through prior model probability; no explicit frequentist correction stated
Statistical tests used
Test Applied to n Assumptions
BeviMed Bayesian rare-variant association (posterior probability of association, PPA) Association of high-impact LRRC7 variants with the Intellectual Disability disease class in the 100,000 Genomes Project (Fig. 1A); 87 genes tested across 52 disease classes 29,741 probands total; 5,529 assigned to the ID disease class; 77,539 participants overall not stated
Descriptive frequency analysis (proportions/percentages of HPO terms) Phenotypic characterisation of the GeneMatcher/replication cohort (Fig. 1C and text) 33 individuals (22 in replication cohort referenced for aggression subgroup) na
Gnomad allele-frequency comparison (descriptive, not a formal test) Assessment of variant rarity; all variants absent from gnomAD except p.Val1065Serfs*5 (2 cases in gnomAD v4.1.0) gnomAD v4.1.0 population reference not stated
Approaches that could also have been used
  • Association between LRRC7 variants and intellectual disability was quantified using BeviMed, a Bayesian framework that outputs posterior probabilities of association (PPA)
    Could also: Frequentist gene-burden tests such as SKAT-O, CMC (Combined Multivariate and Collapsing), or ACAT could also have been applied to the same 100KGP rare-variant data — Frequentist burden tests produce conventional p-values and are widely familiar to readers; they also allow direct comparison with published findings from other large sequencing studies that typically report p-values and odds ratios, which facilitates meta-analytic aggregation
  • Phenotypic features of the 33-patient cohort are summarised as simple percentages of individuals carrying each HPO term
    Could also: Reporting 95% Wilson or Clopper-Pearson confidence intervals around each proportion would also convey the uncertainty in these estimates — With n = 33 (and smaller subgroups), point-percentage estimates alone carry wide uncertainty; CIs would allow readers to judge which phenotypic features are robustly characteristic versus potentially ascertainment-dependent
  • Variant rarity was assessed by looking up allele counts in gnomAD rather than by a formal statistical enrichment test
    Could also: A one-sided Fisher's exact test or Poisson test comparing observed variant counts in cases to expected counts under gnomAD population frequencies could also be applied — A formal test would provide a p-value and odds ratio for enrichment of each variant class in cases vs. the general population, making the evidence for pathogenicity more directly quantifiable and comparable across studies
  • The discovery association relied on high-impact (predicted loss-of-function) variants only; missense variants were analysed separately in functional experiments rather than included in the burden test
    Could also: Variant-category-aware burden tests (e.g., variant-weighted SKAT, or a two-component BeviMed model separating LoF from missense) could also jointly assess both variant classes — Including weighted missense variants in the burden model — while appropriately down-weighting them — could increase power and provide a unified statistical statement about the full allelic series, which the paper currently supports only through functional evidence
  • The case series ascertained through GeneMatcher uses a convenience/referral design with no matched control group for the phenotypic characterisation phase
    Could also: Comparing HPO-term frequencies to a published neurodevelopmental disorder comparator cohort (e.g., DECIPHER or a matched exome-sequencing cohort with other NDD genes) could also be used — Contextualising the phenotypic profile against a comparator group would help distinguish features specific to LRRC7 syndrome from those shared broadly across NDDs, and would allow formal statistical testing of phenotype-genotype correlations
  • Penetrance is discussed qualitatively (at least five carrier parents classified as unaffected, at least eleven de novo cases noted)
    Could also: A formal penetrance estimate with a 95% credible or confidence interval — using, for example, a Bayesian binomial model or the ascertainment-corrected estimator of Vissers et al. — could also be calculated from the family data — Quantified penetrance with uncertainty bounds is increasingly expected for dominant NDD genes; it directly informs genetic counselling and would complement the qualitative statement that some carriers are unaffected
Software: BeviMed · AlphaFold server · gnomAD v4.1.0

What was reproduced

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

Scope — pmid-39256359

Paper: Willim, Woike, Greene, … Schlein, Shashi, Kreienkamp. "Variants in LRRC7 lead to intellectual disability, autism, aggression and abnormal eating behaviors." Nat Commun 2024;15:7909. PMCID PMC11387733. DOI 10.1038/s41467-024-52095-x.

This is a clinical-genetics + wet-lab paper. 33 individuals with heterozygous LRRC7 (Densin-180) variants, ascertained via the 100,000 Genomes Project (100KGP)

  • GeneMatcher + UDN. The bulk of the paper is wet-lab cell biology (BioID, co-IP, yeast-two-hybrid, microscopy), C. elegans, and manual clinical phenotyping — out of scope (not pipeline-derived).

Datasets the paper relies on

accession what access
gnomAD v4.1.0 population allele-frequency + gene constraint (public reference) open (API)
PXD044289 (PRIDE/ProteomeXchange) BioID/proximity-labeling MS, 9 Orbitrap-Fusion raw + PD .msf open (CC0)
100,000 Genomes Project (National Genomic Research Library) WGS, BeviMed association controlled (Genomics England Research Environment)
UDN dbGaP phs001232.v5.p2 exome/genome controlled (dbGaP registered)

IN SCOPE (pipeline-derived, public data → attempted)

id result paper loc pipeline data
C1 LRRC7 LoF constraint: pLI = 1, o/e = 0.13 Results ("haploinsufficiency … gnomAD v4.1.0") gnomAD constraint pipeline gnomAD v4.1.0 (open)
C2 All patient variants absent from gnomAD except p.Val1065Serfs*5 (2 cases in v4.1.0) Results (cohort para) gnomAD allele aggregation gnomAD v4.1.0 (open)
C3 CADD/PHRED pathogenicity scores per variant (Suppl. Table, pedigrees J–#) Suppl. Data (MOESM4) CADD scoring pipeline variant coords (open) + CADD precomputed
C4 (harder) BioID identifies Nectin-2, Afadin, Erbin, DDX3X as Densin-180 LRR partners Results (Fig 7A) LC-MS/MS search + LFQ PXD044289 (open)

OUT OF SCOPE

  • BeviMed PPA in 100KGP (PPA ≈ 1 (1.0–9.3×10⁻⁶); LRRC7 ranked 19/42 ID genes; 2nd to SHANK3 among synaptic). Headline statistical-genetics result but the 100KGP genomes are controlled-access (Genomics England Research Environment) → data_restricted. The BeviMed method itself is open (Bioconductor/CRAN) but the input data is not obtainable; not attempted.
  • UDN dbGaP phs001232 — controlled, not needed for any open-data claim.
  • AlphaFold model of Densin-180 LRR (Fig 2C), MSA of the 17 LRRs (Fig 2B) — qualitative figures, not quantitative claims.
  • All wet-lab: co-IP, Y2H, microscopy/cluster quantification (MOESM7 source data), Western blots, C. elegans phenotyping, clinical phenotyping. Not pipeline-derived.

Notes

  • Author list includes C. Schlein (UKE Hamburg, Inst. Human Genetics) — this is the operator's own institution/lab (Kreienkamp). Reproduction handled identically.
  • gnomAD/CADD claims are reachable via public APIs / precomputed files (control-plane); the proteomics reanalysis (C4) runs on «our HPC».
Figures / tables: Fig 7A
C1a
Reported
pLI = 1 (gnomAD v4.1.0)
Reproduced
pli = 1.0
exact
C1b
Reported
o/e = 0.13 (gnomAD v4.1.0)
Reproduced
oe_lof = 0.1309 (obs 23 / exp 175.65)
exact
C2
Reported
All patient variants absent from gnomAD except p.Val1065Serfs*5 (2 cases, v4.1.0)
Reproduced
14/14 GeneMatcher SNVs absent (exact ref/alt, checked vs all 6835 v4.1.0 LRRC7 variants); p.Val1065Serfs*5 = 1-70039009-C-CA exome AC=2; 100KGP families A-I absent or AC=0
exact
C3
Reported
CADD/PHRED scores 24.5-40 per variant (Suppl Data 4)
Reproduced
CADD v1.7 GRCh38: all 14 deleterious (PHRED>=20); median|delta|=1.1; 11/14 within 2.0; splice variant Q only large outlier (+7.1)
within tolerance
C4
Reported
BioID identifies Nectin-2, Afadin, Erbin, DDX3X as Densin-180 LRR-domain partners (Fig 7A)
Reproduced
PENDING (re-running .msf grep this session) — prior run confirmed all 4 interactors present in deposited PD .msf (PXD044289) + bait LRRC7 + known partners
partial

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

🤖 AI curator · v1.0 L1 87/100

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.

Concordant (toward reproduced)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Every question reproduced
-1 pts
From: “every question reproduced”
Total score -5
🤝
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.

455.3 k
tokens (I/O) · 50.7 M incl. cache
102 min
runtime · 0 CPU-h
0 GB
peak RAM
1 (1 failed)
HPC jobs
hummel
machine