Single-Cell Sequencing of iPSC-Dopamine Neurons Reconstructs Disease Progression and Identifies HDAC4 as a Regulator of Parkinson Cell Phenotypes.
Provisional — an automated or curator check raised a specific concern and points reviewers here. This is NOT a final assessment and not a determination about the authors.
The main result did not reproduce in this reproduction attempt. Where our recomputation produced values that differ from the published ones, those discrepancies are listed below. This is a single automated attempt — not peer review and not a finding of error or misconduct — and differences can also arise from data access, undocumented parameters or the computing environment. The verdict can be contested via “report an error”.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- Nothing in this column.
- 🟡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 deviation was non-trivial in magnitude
- 🔴The central claim did not (fully) hold under reproduction
- 🔴Overall, the reproduction showed a material discrepancy
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
P16 reproduction (ouija = third-party Bayesian pseudotime tool applied to ArrayExpress E-MTAB-7303). DESCRIBED WELL ENOUGH? PARTIALLY. (1) The FASTQ->expression-matrix step is well-described and REPRODUCES: kallisto+tximport on the paper's own plate-1 FASTQs (96 cells) yields a coherent 36552x96 log-TPM matrix with the right biology (TH/STMN2/MAP2 high, ER-stress genes present) -- partial, version-shifted since kallisto v0.42.5 is unobtainable and no per-gene paper values exist to grade against. (2) The designated tool ouija INSTALLS and compiles its Stan model, but is UNRUNNABLE in the only obtainable dependency stack (R4.1.3/rstan2.21.8): every fit allocates pathological memory (187 GiB on the bundled 400x11 example), so no pseudotime could be produced -- env_unresolvable at runtime. (3) The paper's HEADLINE result -- an ouija-derived disease-progression axis on 146 cells / 60 genes -- is NOT independently verifiable: E-MTAB-7303 deposits ONLY raw FASTQ; no processed matrix, no 60-gene marker list, no ouija parameters, no numeric ouija output, and no analysis code were ever published. NOT ATTEMPTED / NOT CHASED (the 20%): aligning all 554 cells; reconstructing the exact 146-cell QC set (criteria underspecified); pinning an older rstan to defeat the ouija memory bug; reproducing the unpublished disease-axis numbers. Drops-are-valid: the central claim rests on un-shipped intermediates -- flagged for human audit, not as fabrication but as non-verifiable-as-published.
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.
-
v1 current initial assessment Score 30assessed: 2026-06-16 ⛓ 4895e7f08a0a
✎ 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.
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
- 2026-06-16
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-16no human curator yet
- Last updated
- 2026-09-19
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: sonnetThe authors test whether high-resolution single-cell transcriptomics of iPSC-derived dopamine neurons carrying the GBA-N370S Parkinson's disease risk variant can exploit cell-to-cell heterogeneity to reconstruct a pseudotemporal axis of disease progression and reveal upstream regulators/therapeutic targets underlying PD-related cellular phenotypes.
- ★ Single-cell transcriptomic analysis of GBA-N370S iPSC-derived dopamine neurons identifies a progressive axis of gene expression variation leading to endoplasmic reticulum stress. finding
- ★ Pseudotime analysis of genes differentially expressed along this axis identifies HDAC4 as an upstream transcriptional repressor regulating early disease-associated gene downregulation. mechanism
- ★ HDAC4 is mislocalized to the nucleus in PD GBA-N370S iPSC-derived dopamine neurons. finding
- ★ Treatment with HDAC4-modulating compounds upregulates genes early in the disease-expression axis and corrects PD-related cellular phenotypes. finding
- ★ Single-cell RNA-seq stratifies patients with clinically similar presentation, identifying one GBA-N370S patient (GBA3) later reclassified as progressive supranuclear palsy. finding
- Bulk RNA-seq identifies 247 genes differentially expressed between GBA-N370S PD and control iPSC-derived dopamine neurons at 1% FDR. finding
- ★ A core set of 60 functionally convergent genes (52 down, 8 up) is identified at the intersection of bulk and single-cell differential expression analyses and is enriched for known PD genes. finding
- HDAC4 mislocalization and perturbation of the same core DE gene set are also observed in a subset of idiopathic (sporadic) PD patient iPSC-derived dopamine neurons. finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| bulk RNA-seq | FACS-purified TH+ iPSC-derived dopamine neurons (3 controls vs 3 PD GBA-N370S patients) | GBA-N370S genetic variant (disease vs control) | differential gene expression (DESeq2, FDR) | — |
| single-cell RNA-seq (plate-based deep sequencing) | FACS-purified TH+ iPSC-derived dopamine neurons (control and GBA-N370S PD patients) | GBA-N370S genetic variant (disease vs control) | single-cell transcriptomes; PCA, over-dispersion, clustering (SC3), pseudotime (switchde) | — |
| FACS | differentiated iPSC-derived neuronal cultures | none | purification/sorting of TH+ dopamine neurons and cell yield | — |
| qRT-PCR | iPSC-derived dopamine neurons from 3 GBA3 clonal lines, 3 controls, GBA1, GBA2, and GBA4 patients | GBA-N370S genetic variant | expression of RPS12, RPS17, RPS6 (SRP pathway genes) | — |
| qRT-PCR time-course | iPSC-derived dopamine neurons at 22 and 38 days in vitro (DIV) | GBA-N370S genetic variant | expression of HDAC4-regulated genes (TSPAN7, ATP1A3, RTN1, PRKCB) and ER stress genes (ERO1A, FKBP9, PDI) | — |
| Western blot | iPSC-derived dopamine neurons (control vs PD GBA-N370S) | GBA-N370S genetic variant | total HDAC4 protein levels | — |
| Ingenuity Pathway Analysis (bioinformatic upstream regulator analysis) | core 60-gene DE set from iPSC-derived dopamine neuron RNA-seq | none | identification of HDAC4 as upstream repressor | IPA (QIAGEN) |
| phenotypic linkage network analysis | core 60-gene set compared to known PD genes and background gene sets | none | functional similarity enrichment score | — |
- – 247 genes differentially expressed between GBA-N370S PD and control purified dopamine neurons by bulk RNA-seq 247 genes at 1% FDR
- ▲ 143 genes significantly over-dispersed in single-cell data, enriched for the SRP-dependent co-translational protein targeting to membrane pathway, specific to patient GBA3 143 genes (0.6%) at 5% FDR
- – Core set of 60 genes identified as DE across bulk and single-cell analyses and as SC3 discriminating markers 60 genes (52 down, 8 up)
- – 60-gene set shows significantly higher functional similarity within a phenotypic linkage network than background genes p < 2.6e-16
- – 60-gene set shows significant functional similarity enrichment with known PD genes compared to background p = 8.52e-08
- – Downregulation of HDAC4-controlled genes (PRKCB, RTN1, ATP1A3, TSPAN7) at 22 DIV precedes upregulation of ER stress genes (ERO1A, FKBP9, PDI) at 38 DIV along the pseudotemporal axis
- – Total HDAC4 protein levels unchanged between controls and PD GBA-N370S patients, despite downregulation of four HDAC4-regulated target genes
- ▲ HDAC4 mislocalized to the nucleus in PD GBA-N370S iPSC-derived dopamine neurons
- count 247 genes DE (bulk RNA-seq, GBA-N370S PD vs control, 1% FDR)
- count 143 genes (0.6%) over-dispersed (single-cell RNA-seq over-dispersion analysis, 5% FDR)
- count 60 core DE genes (52 down, 8 up) (intersection of bulk DE, single-cell DE, and SC3 marker genes)
- pvalue p < 2.6e-16 (functional similarity of 60-gene set vs background in phenotypic linkage network)
- pvalue p = 8.52e-08 (functional similarity enrichment between 60-gene set and known PD genes)
- count 146 single-cell transcriptomic profiles (single-cell RNA-seq profiles passing quality control)
- count ~35,000-40,000 TH+ neurons (FACS-purified per control/PD GBA-N370S sample)
- count 3 controls and 3 PD GBA-N370S patients (plus 4th patient GBA4 in validation) (study cohort for bulk/single-cell RNA-seq and qRT-PCR validation)
Statistical methods review
Model: opusA 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 combined bulk and plate-based deep single-cell RNA-seq of FACS-purified iPSC-derived dopamine neurons from PD GBA-N370S patients and controls. Differential expression was assessed with DESeq2 (at 1% and 5% FDR) for bulk data and with single-cell methods including over-dispersion analysis, switchde, and SC3 clustering, with a two-sided Wilcoxon signed-rank test used for selected pathway-gene comparisons; a Bayesian nonlinear factor-analysis model inferred a pseudotemporal disease axis over a core 60-gene set. Validation experiments (qRT-PCR) were summarized as mean ± SD with significance thresholds, and functional-similarity enrichment was reported with p-values.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| DESeq2 differential expression | bulk RNA-seq, control vs PD GBA-N370S (Figure 1D, 247 genes at 1% FDR; also 5% FDR after GBA3 removal) | three controls and three PD GBA-N370S patients (bulk) | not stated |
| Over-dispersion analysis (Brennecke et al., 2013) | single-cell RNA-seq, identifying genes varying more than technical variation (Figure 2B; 143 genes at 5% FDR) | 146 single-cell profiles passing QC | not stated |
| Two-sided Wilcoxon signed-rank test | DE of SRP pathway genes between GBA3 and controls/other GBA patients (Figure 2D) | — | not stated |
| switchde (single-cell DE across PC2) | genes DE along the pseudotemporal axis (Figure S4B, 5% FDR) | — | not stated |
| SC3 consensus clustering | clustering single-cell RNA-seq to identify discriminating marker genes | — | na |
| Bayesian nonlinear factor-analysis pseudotime model | re-inferring the disease axis over the core 60-gene set (Figure 3A) | — | not stated |
-
qRT-PCR validation results were summarized as mean ± SD with significance-threshold stars (Figures 2E, 3C, S5C).↳ Could also: Reporting exact p-values alongside an effect-size measure and a 95% confidence interval for each comparison would also be possible. — Exact p-values and confidence intervals convey both the magnitude and the precision of the estimate, which is often informative when group sizes are small.
-
Pathway-gene comparisons between GBA3 and other lines used a two-sided Wilcoxon signed-rank test (Figure 2D).↳ Could also: A Mann-Whitney U (rank-sum) test or a mixed-effects model accounting for cell-within-patient structure could also be applied. — The rank-sum test fits independent (unpaired) groups, and a mixed-effects model can account for multiple cells sampled per patient, which would address pseudoreplication when cells share a donor.
-
The validation qRT-PCR experiments involve several genes and group comparisons each reported against significance thresholds.↳ Could also: A single ANOVA with a post-hoc multiple-comparison correction (e.g., Tukey HSD or Bonferroni) could also be used across these comparisons. — A unified model with post-hoc correction controls the family-wise error rate across the related comparisons within an experiment.
-
Differential expression and dispersion analyses applied FDR control at fixed thresholds (1% or 5%).↳ Could also: Reporting effect-size estimates (e.g., log2 fold changes with shrinkage and their standard errors) alongside the FDR-adjusted values is another standard option. — Pairing adjusted significance with shrunken effect sizes helps distinguish statistically detectable from biologically substantial changes.
-
The pseudotemporal disease axis was inferred with a single Bayesian factor-analysis model on the core 60-gene set.↳ Could also: Reporting the trajectory alongside an alternative trajectory-inference method (e.g., diffusion pseudotime or Monocle) as a cross-check is also common practice. — Comparing independent trajectory algorithms can demonstrate robustness of the inferred ordering to methodological choices.
-
Sample size was described by the number of patient lines and cells without a formal power analysis.↳ Could also: A stated power or sensitivity analysis, or a description of the rationale for the number of donors and cells, could also accompany the design. — An explicit power statement helps readers gauge the resolution of the comparisons, particularly given the small number of donor lines.
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.
-
The 60-gene disease-axis set shows significant functional similarity among its members versus background in a phenotypic linkage networkother human ipsc-derived dopamine neuron up 2018×1papers★ This paper is the founder (earliest)
-
247 genes are differentially expressed between PD GBA-N370S and control bulk dopamine neurons, enriched for neuronal development and synaptic functionRNA-seq human ipsc-derived dopamine neuron mixed 2018×1papers★ This paper is the founder (earliest)
-
HDAC4-regulated genes (PRKCB, RTN1, ATP1A3, TSPAN7) are downregulated early along the disease pseudotime axis, preceding late upregulation of ER-stress genes (ERO1A, FKBP9, PDI) in PD GBA-N370S neuronsscRNA-seq human ipsc-derived dopamine neuron mixed 2018×1papers★ This paper is the founder (earliest)
-
A core 60-gene set (52 down, 8 up) defines the control-to-disease pseudotemporal axis in single-cell dopamine neuron transcriptomesscRNA-seq human ipsc-derived dopamine neuron mixed 2018×1papers★ This paper is the founder (earliest)
-
143 over-dispersed genes (0.6%) drive SRP-pathway transcriptional variation specific to GBA-N370S dopamine neuronsscRNA-seq human ipsc-derived dopamine neuron up 2018×1papers★ This paper is the founder (earliest)
-
Total HDAC4 protein levels are unchanged between control and PD GBA-N370S dopamine neurons (while four HDAC4-regulated genes remain downregulated)western-blot human ipsc-derived dopamine neuron none 2018×1papers★ This paper is the founder (earliest)
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.
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.
Downstream reach in the literature
1 downstream papers · 1 datasetsHow widely the datasets deposited by this paper are reused across the whole literature (Europe PMC), beyond our assessed set. This is a factual dependency map — reusing a public dataset is normal, good science. It is not a judgement on the downstream papers; the only verdict here is this paper's own, with its cited rationale.
- Enhancing mitochondrial proteolysis alleviates alpha... 2024 · 12 cites
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-30503143
Paper: Lang, Campbell, et al. (2019) Single-Cell Sequencing of iPSC-Dopamine Neurons Reconstructs Disease Progression and Identifies HDAC4 as a Regulator of Parkinson Cell Phenotypes. Cell Stem Cell. PMID 30503143 / PMC6327112.
Designated code (brief): https://github.com/kieranrcampbell/ouija — a third-party Bayesian single-cell pseudotime tool (Campbell & Yau). This is a P16 reproduction: applying an existing third-party tool to the paper's data is equally valid. It is NOT the authors' full analysis repo (none was deposited).
Data: ArrayExpress E-MTAB-7303 — 554 iPSC-derived dopamine neurons (+14 blanks), Smart-seq2, plate-based, paired-end. Only raw FASTQ on ENA + IDF/ SDRF are deposited; NO processed expression matrix is available.
Reported computational pipeline (STAR Methods)
FASTQ → TrimGalore v0.4.1 (default) → HISAT2 (BAM) / Kallisto v0.42.5 quant vs GRCh38 transcriptome → tximport 1.4.0 (transcript→gene) → QC (plates 3–6 removed) → 146 cells → switchde pseudotime → ouija refined trajectory on a 60-gene core set → disease axis → HDAC4 identified as upstream regulator.
In scope (pipeline-derived, attemptable)
| # | Result | Pipeline | Feasibility |
|---|---|---|---|
| C1 | ouija (the third-party tool) installs & reconstructs a pseudotime | ouija/rstan | HIGH — bundled example_gex, deterministic-ish MAP |
| C2 | FASTQ→expression-matrix step runs on E-MTAB-7303 (TrimGalore/Kallisto/tximport) | kallisto+tximport | MEDIUM — old kallisto v0.42.5, can run modern equivalent on a slice |
| C3 | end-to-end ouija pseudotime on the paper's own data (P16 applicability) | full | MEDIUM — possible, but no published value to compare |
Out of scope / NOT gradeable 1:1 (the deliberate 20% not chased)
- Paper's specific ouija pseudotime / disease axis / switch orderings: the
paper reports no marker gene list, no ouija parameters, and NO numeric ouija
output. There is no published value to compare against →
no_expected_result. - The 60-gene core set: not listed in paper or any deposited file →
docs_insufficient. - HDAC4 as upstream regulator: interpretive/wet-lab-validated downstream conclusion, not a directly-gradeable pipeline number → out of scope.
- Exact 146-cell QC set: QC criteria ("plates 3–6 removed") underspecified; plate→cell mapping not cleanly in SDRF → not reliably reconstructable.
- HISAT2 BAMs, ERCC normalization details: under-described.
Honest assessment
The third-party tool (ouija) is reproducible (C1) and applies to the paper's data (C2/C3). But the paper's own headline computational result cannot be verified 1:1 because nothing numeric was reported and no matrix/markers/analysis-code were deposited — only raw FASTQ. This gap is itself the auditable finding.
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
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.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
The reproducible step — FASTQ→kallisto→tximport — works and yields a biologically coherent 36552×96 plate-1 matrix (TH/STMN2/MAP2 high, ER-stress genes present), so there is no demonstrated factual error. But the paper's headline ouija-derived disease-progression axis (146 cells, 60-gene core set, HDAC4) is non-verifiable-as-published: E-MTAB-7303 deposits only raw FASTQ, with no processed matrix, marker list, ouija parameters, numeric output, or analysis code, and the tool itself is unrunnable in the obtainable environment. The gap is primarily on the authors' side (un-shipped intermediates) plus an underspecified QC cohort on our side — not fabrication, but the central claim rests entirely on artifacts that were never shared, so q5/q7/q8 are red.
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 [email protected].
Reproduction footprint
claude-opus-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.