Differentiation status determines the effects of IFNγ on the expression of PD-L1 and immunomodulatory genes in melanoma.
The main results reproduced, with only marginal, non-material deviations.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓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
- Every checked point held up.
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 to reproduce. The repo (adelomana/nautholsvik, authors' own) + GEO GSE283655 fully specify a kallisto 0.46.1 -> tximport -> DESeq2 1.26.0 LRT pipeline with a Python non-additive quantifier on the published TPM matrix. TIER A (reproducible from the published GSE283655_DESeq2_TPM_values.tsv alone, no FASTQ): EXACT 1:1 reproduction of the two TPM-derived non-additive gene counts setB=222 and setD=619 (matching the repo's own notebook comments), plus correct directional non-additivity for the paper's highlighted immunomodulatory genes CXCL10 (+2.17), CCL2 (+1.16) and CD274/PD-L1 (+0.71 = 1.64x, in the >150% set). Computed by a faithful pure-stdlib re-implementation on «host» (deterministic, platform-independent; 2.5MB matrix, sha256 recorded). TIER B (full FASTQ->kallisto->DESeq2 re-quantification of the 12 RNA-seq samples, ~180GB, PRJNA1194989) is fully staged and tested (tierB.sbatch + tierB_deseq2.R + metadata.tsv) but NOT executed: it requires «our HPC» and the «our HPC» VPN was gated on human 2FA that did not complete during the session. NOT ATTEMPTED: all wet-lab results (qPCR, Luminex, flow cytometry, western blot, inhibitor/siRNA mechanism experiments) and the secondary 45-patient-derived-cell-line analysis (separate dataset). AUDIT FLAG: paper reports 148 non-additive response genes (Fig 2) while the repo notebook comment states 66 for the identical filter definition; the TPM-only superset (222) was reproduced exactly, and Tier B would regenerate the interaction padj needed to adjudicate 148 vs 66. This is a flag for human review, not a confirmed fabrication.
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 61assessed: 2026-06-15 ⛓ ca3e288745ae
✎ 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
- 2026-06-15
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · v1.0 · run #1 2026-06-15no 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: opusThe paper tests how the differentiation status of melanoma cells (governed by MITF expression) affects their response to IFNγ, particularly the expression of PD-L1 and other immunomodulatory genes.
- ★ Dedifferentiation via MITF knockdown renders 624Mel melanoma cells hypersensitive to IFNγ, causing non-additive (synergistic) upregulation of IFNγ-induced immunoregulatory genes. finding
- ★ IFNγ-stimulated dedifferentiated 624Mel cells show increased PD-L1 protein expression and amplified secretion of CCL2, CXCL10 and IL-10. finding
- ★ The intensified IFNγ-induced PD-L1 expression in dedifferentiated cells is mediated through the canonical JAK-STAT1-IRF1 axis rather than an alternative pathway. mechanism
- ★ In 45 patient-derived melanoma cell lines, dedifferentiated (MITF-low) cells show enhanced inflammatory signaling in response to IFNγ and tend toward higher PD-L1 expression associated with increased IRF1 expression/activity. finding
- ★ The hypersensitivity effect is context-dependent, observed in the 624Mel line and not universal to all melanomas. finding
- A non-additive response analysis framework identified 148 non-additive response genes under concurrent siMITF and IFNγ treatment. method
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| qPCR (quantitative RT-PCR) | 624Mel human melanoma cell line | siRNA MITF knockdown +/- IFNγ (5 ng/mL) | gene expression (2-ΔΔCT, GAPDH normalized) | CFX384 Real-Time System & C1000 Touch Thermal Cycler (Bio-Rad) |
| Bulk RNA sequencing | 624Mel human melanoma cell line | siMITF +/- IFNγ | transcriptome-wide gene expression / differentially expressed genes | NovaSeq 6000 (Illumina); Kallisto/DESeq2 analysis |
| Luminex multiplex cytokine assay | 624Mel human melanoma cell line conditioned supernatants | siCTRL/siMITF +/- IFNγ | secreted cytokine concentrations (18 analytes incl. CCL2, CXCL10, IL-10, PD-L1) | Human Premixed Multi-Analyte Luminex Discovery Assay (R&D Systems LXSAHM-18); Bio-Plex 200 System |
| Flow cytometry (FACS) | 624Mel human melanoma cells | siMITF +/- IFNγ | cell-surface PD-L1 protein (geometric mean) | Attune NxT acoustic focusing cytometer; Anti-PD-L1 ab205921 |
| Western blotting | 624Mel / melanoma cell lysates | siMITF/shMitf, IFNγ, JAK/STAT3/NF-κB inhibitors, STAT1/IRF1 knockdown | PD-L1, MITF, STAT1, pSTAT1(Y701), IRF1 protein levels | Odyssey CLx Imaging System (LI-COR); quantified in ImageJ |
| Lentiviral shRNA stable knockdown | YUMM1.7 mouse melanoma cell line (with HEK-293T for packaging) | shMitf vs scrambled shRNA | stable Mitf-low cell lines | pLV[shRNA]-EGFP:T2A:Puro vectors (Vectorbuilder); puromycin selection |
| Bioinformatic analysis of published RNA-seq | 45 wild-type patient-derived melanoma cell lines (GSE154996) | +/- IFNγ (5 ng/mL, 6 h) | PD-L1/MITF mRNA, IFNγ response score, correlations, GSEA | — |
| Small molecule pathway inhibition | 624Mel human melanoma cells | JAK inhibitor I (300/900 nM), Stattic STAT3 inhibitor (1 µM), QNZ NF-κB inhibitor (100 nM) | PD-L1 mRNA/protein (qPCR, western blot) | — |
- ▲ Dedifferentiation makes 624Mel cells hypersensitive to IFNγ, producing non-additive upregulation of IFNγ-induced genes
- ▲ Increased PD-L1 protein expression on cell membranes of IFNγ-treated dedifferentiated 624Mel cells
- ▲ Amplified secretion of CCL2, CXCL10 and IL-10 in IFNγ-stimulated dedifferentiated cells
- – Identification of 148 non-additive response genes under combined siMITF + IFNγ treatment 148 genes
- – Filtering yielded a set of 1,129 response genes used for non-additive analysis 1,129 genes
- ▲ Dedifferentiated patient-derived cell lines show enhanced inflammatory IFNγ signaling and a trend toward higher PD-L1 expression linked to IRF1
- count 45 wild type melanoma cell lines (patient-derived lines analyzed from GSE154996 (originally n=58))
- count 1,129 response genes (DEGs passing expression/fold-change filter for non-additive analysis)
- count 148 non-additive response genes (concurrent siMITF and IFNγ treatment)
- other abs log2 FC > 1 (effect-size threshold for non-additive response genes)
- pvalue adjusted P < 0.05 (Benjamini–Hochberg α = 0.1) (DESeq2 differential expression significance threshold)
- other TPM < 3 excluded; max expression < 2 TPM filtered (expression filtering thresholds for RNA-seq analysis)
Statistical methods review
Model: sonnetA 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 used a 2×2 in vitro design (siCTRL/siMITF × ±IFNγ) in 624Mel melanoma cells with at least three independent biological replicates, assessed by qPCR, western blot, flow cytometry, Luminex multiplex, and bulk RNA sequencing. Differential gene expression was tested with DESeq2 (Benjamini–Hochberg FDR α = 0.1), and a non-additive response framework using DESeq2 likelihood-ratio tests identified synergistic gene-expression changes. A bioinformatic replication arm analyzed 45 patient-derived melanoma cell lines using Spearman correlation, PCA, and GSEA.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Unpaired Student's t-test (direction not confirmed; statistical methods section is truncated in provided text) | Pairwise comparisons of qPCR, flow cytometry, western blot, and Luminex outcomes across experimental conditions | ≥3 independent biological replicates per group; exact per-experiment n not stated in available text | not stated |
| DESeq2 Wald test | Differential gene expression analysis across experimental conditions (RNA-seq, 624Mel cells) | ≥3 biological replicates implied; exact n not stated | not stated |
| Likelihood-ratio test (DESeq2) | Non-additive response analysis: identification of genes with significant siMITF × IFNγ interaction term | ≥3 biological replicates implied; exact n not stated | not stated |
| Two-tailed Spearman correlation | Association between PD-L1 mRNA and IFNγ response score, and between PD-L1 mRNA and MITF mRNA, across 45 patient-derived cell lines | n = 45 cell lines | not stated |
| Simple linear regression | Association between PD-L1 mRNA and IFNγ response or MITF mRNA expression across 45 patient-derived cell lines | n = 45 cell lines | not stated |
| Permutation test (gene set permutation mode, 1,000 permutations) | GSEA enrichment analyses (Tsoi differentiation gene sets; Hallmark and C3 TFT collections); gene ontology biological process and transcription factor target analyses | 1,000 permutations; sample n not stated for GSEA | not stated |
-
Multiple pairwise unpaired t-tests were applied to compare conditions within a 2×2 factorial design (siCTRL/siMITF × ±IFNγ)↳ Could also: Two-way ANOVA with a post-hoc correction (e.g., Tukey HSD or Šidák) — A two-way ANOVA model jointly estimates main effects and an interaction term within a single test family, providing explicit family-wise error rate control and a direct statistical test for the synergy the paper describes, without requiring a separate non-additive framework
-
Benjamini–Hochberg FDR was applied at the α = 0.1 threshold for initial DEG calling in DESeq2↳ Could also: The more conventional α = 0.05 FDR threshold, or independent hypothesis weighting (IHW) to leverage mean-expression as a covariate — α = 0.05 is the most widely reported threshold and eases cross-study comparison; IHW can increase power by down-weighting lowly expressed genes, which is relevant given the subsequent TPM-based filtering step already applied
-
DESeq2 was used for bulk RNA-seq differential expression analysis↳ Could also: edgeR (quasi-likelihood F-test) or limma-voom — Both are extensively benchmarked for count-based RNA-seq; edgeR and limma-voom show comparable or superior performance at small sample sizes, and running one as a sensitivity check is a common practice to assess robustness of DEG lists
-
Two-tailed Spearman correlation was used to quantify the association between PD-L1 mRNA and IFNγ response or MITF mRNA expression across 45 cell lines↳ Could also: Partial Spearman correlation controlling for a potential confound such as overall inflammatory tone or baseline JAK-STAT pathway activity — A partial correlation would allow estimation of the unique contribution of MITF expression to PD-L1 independently of shared variance driven by broad IFN signaling, directly addressing the paper's mechanistic question
-
GSEA used gene set permutation mode↳ Could also: Sample (phenotype) permutation mode — Sample permutation preserves the gene-gene correlation structure of the ranked list and is generally preferred when group sizes are sufficient (n ≥ 7 per group); gene set permutation is an accepted practical alternative for small-n cell-line studies, as explicitly noted in the GSEA documentation
-
qPCR expression was normalized to a single housekeeping gene (GAPDH) using the 2-ΔΔCT method↳ Could also: Normalization to the geometric mean of two or more validated reference genes (e.g., GAPDH + TBP or GAPDH + ACTB) — Multi-reference normalization reduces bias from any single reference gene whose expression may shift under inflammatory stimulation; MIQE guidelines recommend confirming reference gene stability under the specific experimental conditions applied
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.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-39736644
Title: Differentiation status determines the effects of IFNγ on the expression of PD-L1 and immunomodulatory genes in melanoma. PMID 39736644 · PMC11687009 · DOI 10.1186/s12964-024-01963-6 (Cell Commun Signal 2024) Code: github.com/adelomana/nautholsvik (authors' own; HEAD) Data: GEO GSE283655 — 12 samples, 624Mel melanoma, 2×2 design: siCTRL / siMITF (MITF knockdown) × ±IFNγ, 3 replicates each (GSM8668948–8668959, SRA SRX26995276–26995287, platform GPL24676 NovaSeq 6000).
Pipeline (from Methods + repo)
Trimmomatic 0.39 (default) → kallisto 0.46.1 (Ensembl GRCh38 transcriptome;
repo's biomaRt t2g mapping pinned to Ensembl v96) → tximport →
DESeq2 1.26.0 likelihood-ratio tests for three pairwise transitions
(A0→A1 = IFN; A0→B0 = siMITF; A0→B1 = siMITF+IFN) and a genotype:treatment
interaction model (LRT, reduced = ~genotype+treatment; BH α=0.1, padj<0.1).
A Python notebook (non-additive effects quantifier.ipynb) then computes, on the
published TPM matrix, the per-gene non-additive effect
log2((siMITF+IFN obs)+1) − log2((expected = siCTRL + ΔIFN + ΔsiMITF)+1) and
classifies/counts genes.
Conditions ↔ notebook labels: siCTRL=control(A), siMITF=ko(B); wo_IFN=control(0), with_IFN=ifn(1). PD-L1 = CD274 = ENSG00000120217, MITF = ENSG00000187098.
IN SCOPE (pipeline-derived computational results)
Tier A — reproducible from the PUBLISHED GEO TPM matrix alone (no FASTQ)
The GEO supplementary file GSE283655_DESeq2_TPM_values.tsv.gz IS the exact input
of the non-additive notebook. These counts depend ONLY on TPM (median per
condition, low-expr filter max_TPM≥2, log2FC obs/exp), not on DESeq2 p-values:
- setB genes with |log2FC obs/exp| > 1 → repo comment "222"
- setD genes with |log2FC obs/exp| > log2(1.5) → repo comment "619"
- per-gene non-additive log2FC for key genes (CD274/PD-L1, CCL2, CXCL10, IL10, MITF). This faithfully reproduces the authors' non-additive quantification step.
Tier B — needs full re-quantification (FASTQ→kallisto→tximport→DESeq2)
- transition DEG counts (padj<0.1): A0→A1 (IFN) "4414"; A0→B0 (siMITF) "9116"; interaction "700"; siMITF volcano DEG (FC-filtered) "778" — all from repo comments.
- paper headline numbers: 148 non-additive response genes (Fig 2; interaction
padj<0.05 & |log2FC obs/exp|>1), 1,129 response genes (filtered), 36 IFN
responders (Fig 2B). These require the interaction
padj, hence Tier B.
OUT OF SCOPE (wet-lab / manual — not attempted)
qPCR, Luminex cytokine secretion (Fig 3), flow cytometry & western blot of PD-L1, JAK/NF-κB/STAT3 small-molecule inhibitor and STAT1/IRF1 siRNA experiments. These are bench assays, not bioinformatic pipeline outputs.
SECONDARY / OPTIONAL (separate dataset, not GSE283655)
The "45 patient-derived melanoma cell lines ±IFNγ" generalisability analysis uses a different public RNA-seq dataset (Tsoi/Graeber-type panel), not deposited here. Lower priority; attempt only if Tier A+B complete.
Key audit flag (possible discrepancy, to verify by reproduction)
Repo code comments give the non-additive-significant set (interaction padj<0.05 & |log2FC|>1) as 66 genes; the PAPER reports 148 for the same definition (Fig 2). Repo comments may be stale vs the published run. Tier B will produce the actual reproduced count to compare against both.
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.
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-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.