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

Pleiotropic effects of MORC2 derive from its epigenetic signature

· 2025
PubMed 40302207 ↗ pmid-40302207
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 🟡
Input / endpoint not comparable 1:1
+1 pts
From: Q1 · Data identity 🔴
Total score +8
✓ 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). Method is described well (minfi+limma EWAS -> episignature/DMPs; e1071 SVM on 220 CpGs; DESeq2 DEGs; OUTRIDER/PROTRIDER outliers), but the paper deposits NOTHING publicly: data availability is 'available on request', NCBI pubmed->gds/->sra return 0 links, no GEO/SRA/ArrayExpress/PRIDE/Zenodo accession, and there is no analysis-code repository. The open-access supplement ships only derived OUTPUT tables (DMPs/DEGs/protein outliers) and sample metadata, not the per-sample EPIC beta / RNA-seq count / proteomics intensity matrices needed to run any pipeline. So 0 of 6 in-scope pipeline results are reproducible. I did NOT fabricate a run: 0 «our HPC» jobs. What I COULD verify on the supplement (provisional, not pipeline reproduction): cohort N=53 matches exactly; the Leigh/mito EWAS block contains exactly the reported 80 Bonferroni-significant CpGs; the episignature is predominantly hypermethylation (711/770); and the DMP table's multiple-testing columns are arithmetically self-consistent (Bonferroni = P x ~6.3e5). These are text<->supplement concordance + internal-consistency checks (a fabrication screen, no red flags), NOT independent recomputations. Not attempted: all wet-lab steps and every pipeline result (no inputs). To reproduce, the authors' raw matrices must be obtained 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
    assessed: 2026-06-18 ⛓ 22ea7138af84
✎ 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
2026-06-18
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · 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 molecular mechanism linking MORC2 gain-of-function pathogenic variants to their wide range of phenotypes (pleiotropy and phenotypic heterogeneity) has remained elusive, and the paper tests whether a MORC2-specific epigenetic (DNA methylation) signature underlies this pleiotropy.

Core claims
  • A MORC2-specific DNA methylation episignature exists that is universal across all MORC2-associated phenotypes and conserved across blood and fibroblast tissue finding
  • The episignature consists mainly of DNA hypermethylation in promoter regions, leading to transcriptional repression of target genes and a MORC2-specific RNA signature mechanism
  • Concomitant downregulation of ERCC8, NDUFAF2 and FKTN at different levels mirrors the variable biochemical defects and clinical manifestations observed in MORC2 patients finding
  • Silencing of NDUFAF2 accounts for the Leigh syndrome manifestation in MORC2 patients mechanism
  • Dysmorphic features in MORC2 patients are due to repression of ERCC8 mechanism
  • Methylation level of the bidirectional ERCC8-NDUFAF2 promoter correlates with disease severity, being higher in Leigh syndrome than in CMT finding
  • An SVM classifier trained on 220 episignature CpG sites can serve as a functional diagnostic assay for MORC2 variants of uncertain significance resource
  • Epigenetic variation may underlie pleiotropy in other Mendelian disorders beyond MORC2 finding
Experimental setups
Assay System Perturbation Readout Platform
DNA methylation array (EWAS) blood, patient-derived MORC2 pathogenic variant (heterozygous missense) vs healthy control differentially methylated positions (beta/M-values) Infinium Methylation EPIC v1.0 and v2.0 BeadChip; minfi (R); limma
DNA methylation array (EWAS) fibroblasts, patient-derived MORC2 pathogenic variant vs control fibroblast lines differentially methylated positions Infinium Methylation EPIC BeadChip; minfi (R); limma
SVM classification on DNA methylation blood and fibroblast samples MORC2 disease status vs controls/other genetic disorders classification probability score (episignature positive/negative) e1071 package (R); Caret filterVarImp
Phenotype-specific EWAS fibroblast (Leigh syndrome-enriched) and blood (CMT-enriched) patient samples Leigh syndrome/mitochondrial disease vs CMT phenotype differentially methylated CpG sites between phenotypes limma package
RNA sequencing patient-derived skin fibroblasts MORC2 pathogenic variant transcriptome, gene expression levels Illumina HiSeq2500/HiSeq4000; STAR v2.7.0a; DROP pipeline v1.3.4
Proteomics patient-derived skin fibroblasts (n=13 patients) MORC2 pathogenic variant protein abundance
White blood cell type deconvolution blood none estimated proportions of white blood cell types Houseman method
Epigenetic age estimation blood and fibroblast none predicted biological age from methylation data Horvath method via methylclock package
Key results
  • 631,142 CpG sites across 512 samples (481 blood, 31 fibroblast) passed QC filtering for downstream methylation analysis
  • EWAS discovery in blood identified CpG sites significantly differentially methylated between MORC2 patients and controls at Bonferroni-adjusted P < 0.05 n=760 CpG sites
  • 220 CpG sites selected by feature importance achieved classifier performance with area under the curve greater than 0.99 AUC > 0.99
  • Phenotype-specific EWAS comparing Leigh syndrome to CMT/control samples identified significant CpG sites at Bonferroni-adjusted P < 0.05 n=80 CpG sites
  • ERCC8-NDUFAF2 bidirectional promoter methylation level was higher in Leigh syndrome patients than in CMT patients, correlating with disease severity
Key statistics
  • count 631142 CpG sites; 512 samples (n=481 blood, n=31 fibroblast) (post-QC DNA methylation dataset used for downstream analysis)
  • pvalue Bonferroni-adjusted P < 0.05 (n = 760 CpG) (differentially methylated positions in MORC2 vs control blood EWAS discovery)
  • other AUC > 0.99 (feature importance threshold for selecting 220 CpG sites for SVM classifier)
  • pvalue Bonferroni-adjusted P < 0.05 (n = 80 CpG sites) (phenotype-specific EWAS comparing Leigh syndrome vs CMT/control)
  • count 53 MORC2 genetically diagnosed patients (39 new, 14 previously reported) (study cohort size)
  • count 112 total patients (73 previously reported, 39 new) (phenotypic distribution across MORC2-related disorders)
  • count 47 DNA methylation data points from MORC2 patients (n=35 blood, n=12 fibroblast) (methylation profiling across 45 patients)
  • count 43 MORC2 episignature-positive samples: Leigh syndrome (n=13), mitochondrial disease (n=6), CMT (n=24) (phenotype breakdown used for phenotype-specific episignature discovery)

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 performed epigenome-wide association studies (EWAS) using linear regression (limma package) on blood and fibroblast DNA methylation arrays (EPIC v1.0/v2.0) from MORC2 patients versus controls, with Bonferroni correction across ~631,000 CpG sites to identify differentially methylated positions. A support vector machine (SVM) classifier was then trained on a curated subset of CpG sites to discriminate MORC2 patients from controls and to distinguish phenotypic subgroups, with 10-fold or 5-fold cross-validation. Complementary transcriptome (RNA-seq via STAR/DROP pipeline) and proteome analyses were performed on fibroblast samples from a subset of patients; the statistical framework for those modalities is not fully captured in the available text excerpt.

Replicationbiological Sample size53 MORC2 patients enrolled; 47 DNA methylation data points available (35 blood, 12 fibroblast); 12 patients for RNA-seq, 13 for proteomics; 80:20 train-test split used for classifier development GroupsMORC2 patients vs healthy controls; Leigh syndrome vs CMT phenotypic subgroups Pairingunpaired Randomization/blindingnot stated Dispersionunclear Multiplicity correctionBonferroni correction as primary significance threshold; FDR (-log FDR × |logFC|) used as a probe-ranking score for feature selection, not as a standalone significance criterion
Statistical tests used
Test Applied to n Assumptions
Linear regression on M-values (limma package) EWAS discovery of MORC2-specific episignature in blood DNA methylation Training: 10 MORC2 patients vs 46 healthy controls (from 13 patients / 58 controls before 80:20 split) not stated
Linear regression on M-values (limma package) EWAS discovery of phenotype-specific episignature comparing Leigh syndrome vs CMT/controls 10 Leigh syndrome patients vs 32 samples (16 CMT patients + 16 controls) not stated
Support vector machine (SVM), linear kernel, 10-fold cross-validation (e1071 package) Diagnostic classifier discriminating MORC2 patients from controls using 220-CpG episignature Training: 10 MORC2 + 46 controls; Test: 3 MORC2 + 12 controls + 368 other genetic disorders; Fibroblast test: 6 MORC2 + 18 controls/diseased na
Support vector machine (SVM), linear kernel, 5-fold cross-validation (e1071 package) Phenotype classifier discriminating Leigh syndrome/mitochondrial disease vs CMT using 80-CpG set Training: 10 Leigh + 32 (CMT+controls); Test: 9 Leigh/mito + 11 CMT; additional cohort n=449 na
Multidimensional scaling (MDS) on Euclidean distances of M-values Unsupervised visualization of sample clustering using episignature CpG sites null na
Approaches that could also have been used
  • Bonferroni correction was applied genome-wide across ~631,000 CpG sites to define significance in both EWAS analyses
    Could also: Benjamini-Hochberg (BH) false discovery rate (FDR) correction could also have been applied as the primary significance criterion — BH-FDR is widely used in EWAS/GWAS contexts because Bonferroni assumes independence of tests and is therefore highly conservative when neighboring CpGs are correlated due to linkage disequilibrium; FDR control can increase sensitivity for discovery while maintaining a specified error rate, which may be desirable when the goal is to build a feature set for a downstream classifier
  • An 80:20 single random train-test split was used to partition patients and controls for classifier discovery and initial performance evaluation
    Could also: Nested cross-validation (outer loop for performance estimation, inner loop for feature selection and hyperparameter tuning) could also have been used — With a small patient group (13 MORC2 patients before splitting), a single held-out split produces performance estimates with high variance; nested CV uses all available labeled data for both selection and evaluation, yielding less biased and more stable estimates of generalization performance
  • Tissue type (blood vs fibroblast) was included as a linear covariate in the phenotype-specific EWAS to control for tissue confounding, while most Leigh patients contributed fibroblast data and most CMT patients contributed blood data
    Could also: Stratified analysis within tissue type, or tissue-matched case-control comparison, could also have been used — Including tissue as an additive covariate assumes a constant, phenotype-independent tissue effect; when tissue and phenotype are substantially confounded (as here), a stratified or matched approach avoids residual tissue-phenotype collinearity and may separate tissue effects from disease effects more cleanly
  • A support vector machine (SVM) with a linear kernel was chosen as the diagnostic classifier
    Could also: Regularized logistic regression (elastic net / LASSO) or random forest could also have been used as classifier models — Elastic net logistic regression provides interpretable, sparse coefficient estimates and explicit built-in feature selection; random forest handles non-linear interactions and yields variable importance measures; both are widely benchmarked in episignature classification studies and offer complementary perspectives on which CpG sites drive discrimination
  • Sample-level aggregated methylation was computed as the mean Z-score of M-values across a defined set of CpG sites
    Could also: A PCA-derived score or an effect-size-weighted composite could also have been used to summarize multi-CpG methylation per sample — Equal-weight averaging assumes all selected CpGs contribute equally to the signal; weighting by effect size (e.g., logFC) emphasizes more discriminating sites, while the first principal component captures the dominant axis of co-variation across CpGs and may more robustly reflect the underlying biological signal
  • Quantile normalization was applied to harmonize EPIC v1.0 and EPIC v2.0 array data, with array version subsequently included as a covariate in regression models
    Could also: ComBat batch correction or functional normalization (FunNorm, available in minfi) could also have been applied to explicitly model array-version batch effects — ComBat estimates and removes batch-specific offsets empirically, which may reduce residual technical variance more completely than a regression covariate; FunNorm uses Illumina control probes for between-array normalization and is designed specifically for Infinium methylation arrays, making it a common alternative in multi-batch EPIC studies
Software: R/minfi R 4.3.1 · R/limma · R/Caret (filterVarImp) · R/e1071 (SVM) · R/methylclock (Horvath epigenetic clock) · STAR (RNA-seq alignment) 2.7.0a · DROP pipeline (RNA-seq QC and read counting) 1.3.4 · R/IlluminaHumanMethylationEPICanno.ilm10b4.hg19

What was reproduced

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

Scope — pmid-40302207

Title: Pleiotropic effects of MORC2 derive from its epigenetic signature Journal: Brain (2025/2026), DOI 10.1093/brain/awaf159, PMCID PMC12782172 (open access). Authors include: Schlein C («host» operator is a co-author).

What the paper does

Multi-omic study of 53 MORC2 patients: blood + fibroblast DNA methylation (Illumina Infinium EPIC v1.0/v2.0 arrays), RNA-seq (fibroblast), proteomics (fibroblast), and HPO phenotypes. Identifies a MORC2-specific DNA-methylation episignature (mainly promoter hypermethylation), a derived RNA signature, and links repression of ERCC8 / NDUFAF2 / FKTN to phenotype.

Pipelines named in Methods (would be in scope if data were available)

Result Pipeline / tool Input needed
DMPs / episignature (EWAS) minfi preprocessing (R 4.3.1) → limma per-sample EPIC beta/M-value matrix (IDAT)
Diagnostic classifier e1071 SVM on 220 CpG M-values, 10-fold CV same beta matrix
Phenotype (Leigh/mito) EWAS, n=80 CpG limma same
DEGs DESeq2 1.34.0 RNA-seq count matrix
Expression outliers OUTRIDER (autoencoder), cohort n=877 RNA-seq count matrix
Protein outliers PROTRIDER, 6749 proteins proteomics intensity matrix

In scope vs out of scope

  • Out of scope — not attempted (wet-lab / manual): sample collection, bisulfite conversion, array hybridisation, clinical phenotyping, variant calling, validation experiments.
  • In scope in principle (pipeline-derived): all six rows above.
  • In scope and ACTUALLY attempted: none can be run. See blocker below.

Reproduction blocker (decisive)

  • No public data deposit. Data-availability statement: "The dataset generated and analysed in the current study is available on request." No GEO/SRA/ArrayExpress/PRIDE accession. NCBI elink pubmed→gds and →sra return 0 LinkSetDb (no linked records).
  • No code repository. Full text contains no GitHub/GitLab/Zenodo/figshare code link (only an annotation-database URL, zwdzwd.github.io/InfiniumAnnotation).
  • Supplement ships derived OUTPUTS only (DMP/DEG/outlier tables + sample metadata), no per-sample beta/count/intensity matrices = no pipeline inputs.

→ Every in-scope pipeline result depends on raw inputs that are on-request only. Outcome: drop, data_restricted. What was done: profile the open-access supplement and run internal-consistency + text↔supplement concordance checks (see AUDIT.md). These are not independent reproductions and are graded as such.

C1_cohort_n
Reported
53 MORC2 patients
Reproduced
ST2 has exactly 53 cohort rows (text<->supplement concordance; not a pipeline repro)
exact
C3_leigh_ewas_80cpg
Reported
n=80 CpG sites Bonferroni<0.05 (mito/Leigh EWAS)
Reproduced
ST4 block2 has exactly 80 Bonferroni<0.05 CpGs (concordance; not a pipeline repro)
exact
C4_episig_hyper
Reported
episignature mainly hypermethylation
Reproduced
ST4 block1 = 711 hyper / 59 hypo of 770 DMPs (concordance; not a pipeline repro)
within tolerance
C9_dmp_bonferroni_internal
Reported
Bonferroni-adjusted p-values (DMP table)
Reproduced
Bonferroni = P x 631142/629984 constant; table internally consistent
within tolerance

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 38/100

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 🟡
Input / endpoint not comparable 1:1
+1 pts
From: Q1 · Data identity 🔴
Total score +8

This is a legitimate DROP (data_restricted): the raw EPIC/RNA-seq/proteomics matrices are 'available on request' with no public deposit and no code repo, so none of the 6 in-scope pipelines could be independently rerun. The limitation sits on the authors'/data-availability side, not in our methodology. What could be checked — cohort N=53, the 80-CpG Leigh block, 711/59 hyper/hypo bias, and internally consistent Bonferroni arithmetic — all match the prose and are self-consistent, a clean fabrication screen with no red flags. Severity of observed deviation is negligible (none found); the central hypermethylation claim is plausibly supported but only at concordance level, hence limited rather than confirmed.

🤝
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.

95.8 k
tokens (I/O) · 4.9 M incl. cache
10 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.