Stemness genes and miR-1247-3p expression associate with clinicopathological parameters and prognosis in lung adenocarcinoma.
The main results reproduced: recomputed values matched the published ones within tolerance.
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
- ✓Reported values are derivable from the shared data
- ✓The central claim held under reproduction
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡The deviation was non-trivial in magnitude
- 🟡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
Described well enough to reproduce the PRIMARY result 1:1. The repo (Sashoss/LUAD_PLOS_data @ 8b6d7e1) ships data + derived outputs but NO authors' analysis code; per brief P16 we re-ran the paper-described pipeline (DESeq2 for TCGA, Limma for GEO; FC=tumor/normal) on the shipped inputs on «our HPC» («infra», conda DESeq2 1.42.0 / limma 3.58.1 / R 4.3). TCGA primary cohort (gene + miRNA DESeq2) reproduces EXACTLY: Pearson cor of log2FC = 1.000 over the whole transcriptome (34924 genes) and miRNA set (1359); all headline stemness genes (ORC1L 2.741, KIF20A 3.144, DLGAP5 3.671, RAB3B 4.242, CENPI 2.740) match Table 1 to 3 decimals, and the headline miR-1247-3p downregulation is exact (-2.325). The GEO validation cohorts (Limma) reproduce direction and ranking but NOT magnitude 1:1 (GSE31210 cor=0.9975 but 1.45x smaller; GSE40419 cor0.70-0.75, inconsistent) because the authors' microarray preprocessing/probe-collapsing is undocumented (no code shipped) - this is the hard ~20% and was not chased further. NOT attempted (out of scope): univariate/multivariate Cox survival (clinical survival not shipped as a runnable table), TIMER2 immune infiltration (external web tool output), miRWalk target prediction (external web DB output). One audit note: an early fabrication suspicion for ORC1L was OUR gene-ID error (ENSG00000113742=CPEB4 not ORC1; correct ORC1=ENSG00000085840) and was RESOLVED - ORC1L reproduces exactly. Net: the paper's core differential-expression claims are faithfully reproducible from its own shipped data via the stated standard pipeline.
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 90assessed: 2026-06-15 ⛓ 093e294c34f8
✎ 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 · claude (ai-curator room) · v1.0 · run #1 2026-06-15no 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: sonnetLung adenocarcinoma's high mortality is linked to acquisition of stem-like cancer cells that help tumors regulate immune cell infiltration, so the study tests whether stemness-related genes and miRNAs in LUAD associate with immune/accessory cell infiltration and patient survival.
- ★ Stemness-related genes ORC1L, KIF20A, and DLGAP5 are upregulated in LUAD and correlate with altered immune cell infiltration (Th2, MDSC, HSC) and reduced patient survival. finding
- ★ hsa-mir-1247-3p, downregulated in LUAD, together with its targets SLC24A4 (tumor suppressor, downregulated) and RAB3B/HJURP (oncogenes, upregulated), is a key regulator of LUAD patient survival. finding
- ★ MDSC infiltration modulation is independently associated with LUAD patient survival via a mechanism distinct from hsa-mir-1247-3p. finding
- 109 stemness-related genes derived via ssGSEA (Miranda et al. 2019) were used to screen for stemness markers differentially expressed in LUAD. method
- ★ hsa-mir-1247-3p and its associated gene targets may offer a promising biomarker/therapeutic avenue to enhance LUAD patient survivability. resource
- RAB3B's 3'UTR is targeted by five differentially expressed miRNAs, with hsa-mir-1247-3p the only one meeting stringent DeSeq2 and Limma cutoffs. finding
- ★ Higher HSC infiltration is associated with improved LUAD patient survival, while higher Th2 and MDSC infiltration are associated with reduced survival. finding
- Cancer stage (II, III, IV vs I) and increased patient age are significantly associated with decreased LUAD patient survival. finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Bulk RNA-seq differential expression (DESeq2) | TCGA LUAD tumor (n=537) vs normal (n=59) tissue | none (tumor vs normal comparison) | differentially expressed genes/miRNAs (log2 fold change, adjusted p-value) | DESeq2 |
| Microarray gene expression differential analysis (Limma) | GSE40419 LUAD tumor (n=87) vs normal (n=77) tissue | none | differentially expressed genes (LFC, adjusted p-value) | Limma |
| Microarray gene expression differential analysis (Limma) | GSE31210 LUAD tumor (n=226) vs normal (n=20) tissue | none | differentially expressed genes (LFC, adjusted p-value) | Limma |
| Univariate Cox-proportional hazard survival analysis | TCGA LUAD patient clinical/expression data | none (high vs low expression/infiltration groups) | hazard ratio and p-value for gene/miRNA expression and immune infiltration vs survival | Timer2.0 |
| Computational immune/accessory cell infiltration deconvolution and correlation analysis | TCGA LUAD tumor samples, 32 immune/accessory cell types | none | correlation between gene expression and infiltration rate | Timer2.0 (Timer, TIDE, MCP-counter, quanTIseq, xCell, CIBERSORT, Epic, immunedeconv) |
| In silico miRNA target prediction (3' UTR binding) | human gene 3' UTR sequences / miRNA database | none | predicted and validated miRNA-gene binding interactions | mirWalk, miRTarBase |
| Multivariate Cox-proportional hazard model survival analysis | TCGA LUAD patient data (age, stage, gene/miRNA expression, MDSC/Th2/HSC infiltration) | none | hazard ratio and p-value for combined covariates vs survival | Survival Analysis package in R |
| MDSC infiltration estimation | TCGA LUAD tumor samples | none | MDSC infiltration rate | TIDE |
- ▲ ORC1L, KIF20A, DLGAP5, RAB3B differentially expressed (upregulated) in LUAD across TCGA, GSE40419, GSE31210 with hazard ratio >1 e.g. DLGAP5 LFC=3.671 (TCGA DeSeq2)
- – ORC1L, KIF20A, DLGAP5 positively correlate (>0.6) with Th2 and MDSC infiltration; ORC1L negatively correlates (<-0.6) with HSC infiltration correlation >0.6 / <-0.6
- – High Th2 and MDSC infiltration associated with reduced survival; high HSC infiltration associated with improved survival HR=3.804 (Th2), HR=62.036 (MDSC), HR=0.167 (HSC)
- ▼ hsa-mir-1247-3p downregulated in LUAD; the only one of 5 candidate miRNAs meeting LFC and BH-adjusted p cutoffs targeting RAB3B 3'UTR LFC=-3.151 (Limma), BH p=1.63E-11
- – In multivariate analysis, increased hsa-mir-1247-3p expression associated with loss of survival despite overall downregulation; MDSC infiltration independently associated with loss of survival HR=0.667653 (mir-1247-3p low vs high expr), p=0.02279; HR=0.655898 (MDSC low vs high), p=0.028825
- – No significant correlation between MDSC infiltration and hsa-mir-1247-3p expression Pearson r=-0.05
- – 53 common hsa-mir-1247-3p target genes identified; HJURP and RAB3B consistently upregulated, SLC24A4 consistently downregulated across all three datasets LFC>2 (HJURP, RAB3B); LFC<-2 (SLC24A4)
- ▲ Cancer stage II/III/IV and increased age significantly associated with decreased survival relative to stage I HR=1.94 (stage II), HR=3.17 (stage III), HR=3.63 (stage IV)
- correlation r=-0.05 (MDSC infiltration vs hsa-mir-1247-3p expression, not significant)
- fold_change LFC=3.671 (DESeq2, TCGA) (DLGAP5 upregulation in LUAD tumor vs 59 normal samples)
- fold_change LFC=-3.151 (hsa-mir-1247-3p downregulation (Limma), BH adjusted p=1.63E-11)
- pvalue HR=62.036, p=0, BH adj p=0 (MDSC high vs low infiltration, univariate Cox survival analysis)
- pvalue HR=3.804, p=0.006, BH adj p=0.009 (Th2 high vs low infiltration, univariate Cox survival analysis)
- pvalue HR=0.667653, p=0.02279 (hsa-mir-1247-3p low vs high expression, multivariate Cox model)
- count 109 (stemness-related genes screened from Miranda et al. 2019)
- count 53 (common hsa-mir-1247-3p target genes identified via mirWalk and miRTarBase)
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.
This retrospective bioinformatics study used TCGA RNA-seq data (537 tumor, 59 normal) and two GEO microarray validation datasets to identify stemness-related genes and miRNAs differentially expressed in lung adenocarcinoma. Differential expression was assessed with DESeq2 (RNA-seq) and limma (microarray), both with Benjamini-Hochberg FDR correction at adjusted p ≤ 0.05 and |log2FC| > 2. Univariate and multivariate Cox proportional hazard models were then used to evaluate associations between gene/miRNA expression, computationally predicted immune cell infiltration rates, and patient survival, with results reported as hazard ratios and exact BH-adjusted p-values.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| DESeq2 Wald test | Differential expression of genes and miRNAs in TCGA LUAD tumor vs. normal samples | 537 tumor + 59 normal (TCGA) | not stated |
| limma moderated t-test | Differential expression of genes in GEO microarray datasets GSE40419 and GSE31210 used as validation cohorts | 87 tumor + 77 normal (GSE40419); 226 tumor + 20 normal (GSE31210) | not stated |
| Univariate Cox proportional hazard model | Association between high vs. low gene expression or immune cell infiltration and LUAD patient survival, performed via Timer2.0 | — | not stated |
| Multivariate Cox proportional hazard model | Joint association of miRNA/gene expression, immune infiltration rates, patient age, and cancer stage with patient survival (R Survival package) | — | not stated |
| Pearson correlation | Association between stemness gene expression and computationally predicted infiltration rates of 32 immune/accessory cell types; also reported for MDSC vs. hsa-mir-1247-3p (r = -0.05) | — | not stated |
| Kaplan-Meier survival curve | Visual display of survival probability for high vs. low immune cell infiltration and gene expression groups, obtained from Timer2.0 | — | na |
-
Patient samples were dichotomized into high and low expression (or infiltration) groups before Cox proportional hazard model analysis↳ Could also: Cox regression treating gene/miRNA expression as a continuous variable, or using a restricted cubic spline to model non-linear dose-response — Continuous Cox modeling preserves information across the full expression range, avoids sensitivity to the choice of cutpoint, and is generally more statistically efficient than a median-split dichotomization
-
Pearson correlation was used to assess associations between gene expression and computationally predicted immune cell infiltration rates across 32 cell types↳ Could also: Spearman rank correlation — Spearman correlation does not assume linearity or bivariate normality, assumptions that may not hold for RNA-seq-derived values and computationally estimated infiltration scores
-
Hazard ratios from Cox proportional hazard models were reported with p-values but without confidence intervals↳ Could also: Report 95% confidence intervals alongside each hazard ratio — Confidence intervals directly convey the precision and uncertainty of each hazard ratio estimate, complementing the p-value and making effect magnitude more interpretable
-
Correlations between gene expression and 32 immune cell types were screened using a |r| > 0.6 threshold with nominal p ≤ 0.05, without an explicit multiplicity correction for the 32 cell-type tests↳ Could also: Apply BH-FDR correction across the family of 32 correlation tests per gene — Testing 32 cell types per gene increases the probability of finding at least one spurious correlation at a nominal threshold; a within-family correction would be consistent with the approach applied to the differential expression and survival analysis steps
-
Univariate Cox models were used to pre-screen individual genes and immune cell types for survival associations before entry into the multivariate model↳ Could also: LASSO-penalized Cox regression for simultaneous covariate selection in a single multivariate model — LASSO-penalized Cox regression selects predictors jointly, reducing the potential for bias introduced by sequential univariate pre-screening and handling correlated predictors more explicitly
-
DESeq2 was the sole method used for RNA-seq differential expression in the TCGA dataset↳ Could also: edgeR (quasi-likelihood F-test) or voom/limma applied to RNA-seq count data as a concordance check — DESeq2, edgeR, and voom/limma are all widely validated for RNA-seq differential expression; reporting genes that are significant across two methods is a common strategy for increasing confidence in findings, analogous to the cross-platform validation the paper applied to the microarray data
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-37948380
Paper: Limbu & McCloskey 2023, PLoS One. "Stemness genes and miR-1247-3p expression associate with clinicopathological parameters and prognosis in LUAD." DOI 10.1371/journal.pone.0294171 · PMCID PMC10637681.
Code/Data artifact: https://github.com/Sashoss/LUAD_PLOS_data @ commit
8b6d7e125fdbb20685a32a9246bc7f2f4846784f (2023-09-01). This is a data + derived-results
repository — it ships NO authors' analysis scripts (no .R/.py/.ipynb/README). Per
brief rule P16, reproduction = re-run the standard, paper-described pipeline on the
shipped inputs and compare to the shipped derived outputs + paper claims. This is
fully valid.
Pipeline (from paper Methods)
- Differential expression:
- TCGA gene & miRNA counts → DESeq2.
- GEO microarray/RNA-seq (GSE40419, GSE31210) → Limma.
- DEG cutoff: Benjamini-Hochberg adj p ≤ 0.05 AND |log2FC| > 2. FC = cancer/normal.
- Univariate & multivariate Cox proportional-hazard survival analysis (high vs low expr).
- TIMER2 immune-infiltration correlation; miRWalk target prediction.
Shipped inputs (on «infra», complete, sizes match GitHub)
data_gene.txt(99 MB): TCGA STAR raw gene counts, ENSG ids, 596 samples (537 Primary_Tumor + 59 Solid_Tissue_Normal; condition encoded in column-name prefix).data_mirna.txt(2.5 MB): TCGA miRNA raw counts, 565 samples (519 T + 46 N).GSE40419-GPL11154_withGeneName.csv(17 MB) +GSE40419_annotation.csv(87 T / 77 N).GSE31210-GPL57_withGeneName.csv(53 MB) +GSE31210_annotation.csv(226 T / 20 N).Metadata_LUAD_gene.tsv,Metadata_LUAD_mirna.tsv(clinical/sample metadata).
Shipped derived outputs (comparison ground-truth)
TCGA_diffExp_gene_DeSeq2.csv(full transcriptome),TCGA_diffExp_mirna_DeSeq2.csv.GSE40419_diffExp_Limma.csv,GSE31210_diffExp_Limma.csv,TCGA_diffExp_gene_Limma.csv(latter three restricted to a ~15-gene stemness panel).multivariate_survival_analysisData_Table1/2.csv,Timer2_TCGA_infiltration_data.csv.
IN SCOPE (attempted — clearly specified pipeline, inputs+outputs shipped)
- TCGA DESeq2 gene diff-exp (tumor vs normal) → compare to
TCGA_diffExp_gene_DeSeq2.csv. - TCGA DESeq2 miRNA diff-exp → compare to
TCGA_diffExp_mirna_DeSeq2.csv(key: hsa-mir-1247). - GSE40419 Limma diff-exp → compare to
GSE40419_diffExp_Limma.csv(stemness panel). - GSE31210 Limma diff-exp → compare to
GSE31210_diffExp_Limma.csv(stemness panel).
Key reported claims compared: 3 stemness genes ORC1L/KIF20A/DLGAP5 up in LUAD; hsa-mir-1247-3p down; targets SLC24A4 (down), RAB3B/HJURP (up).
OUT OF SCOPE (80/20 — not attempted, reasons)
- Univariate/multivariate Cox survival (Table 1/2): requires cBioPortal-sourced clinical survival + stage/age not fully shipped in a single runnable table; lower priority than diff-exp; skipped per 80/20.
- TIMER2 immune infiltration: TIMER2 is an external web tool; shipped CSV is its output, not reproducible without re-querying the web service. Out of scope.
- miRWalk target prediction: external web DB output (xlsx), not a runnable pipeline.
- Wet-lab / manual interpretation claims: out of scope by definition.
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 paper's core differential-expression claims reproduce 1:1 from its own shipped TCGA data: whole-transcriptome and whole-miRNA log2FC correlate at 1.000 and every Table 1 headline stemness gene plus miR-1247-3p matches to 3 decimals, so q5/q7 are fully on the authors' side as confirmed. The only deviations sit in the GEO validation cohorts, where Limma magnitudes run ~1.45x low (GSE31210) or inconsistent (GSE40419 RAB3B 3.152→0.494) — direction and ranking still hold (GSE31210 cor=0.9975). This gap is on our-method/availability side, caused by the authors shipping no analysis code and not documenting their microarray preprocessing, not by fabrication (the early ORC1L suspicion was our own gene-ID error, resolved). Overall: a strong primary reproduction with moderate, fully explainable secondary-cohort magnitude deviations → yellow.
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.