Multi-omics analyses identify mannose phosphate isomerase-centered hypoxia-induced angiogenesis signature in colorectal cancer.
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.
- ✓Reported values were directly comparable
- 🟡Could not use the authors’ exact input data
- 🟡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
DESCRIBED WELL ENOUGH + 1:1 on the headline pipeline result. SRGA (https://github.com/LucasLiu20200131/SRGA @6c77e032, authored by first author S. Liu) is the paper's signature-related-gene-analysis tool. (1) On «our HPC» I built an R4.3+Bioconductor-fgsea env on «infra» and ran the repo's shipped worked example end-to-end: it executes and is bit-for-bit deterministic under a fixed seed (R1, within-tol). (2) The paper's headline pipeline number -- '911 unique angiogenesis-related genes' -- is EXACTLY reproduced as the unique-gene dedup of shipped Supplementary Data 4, which is the verbatim SRGA fgsea output table (1255 gene-signature rows -> 911 unique symbols; all rows satisfy |AngScore|>0.95 & FDR<0.05). Column schema matches R/fgsea_calculation.R, so the value is genuinely pipeline-derived, not fabricated (R2, exact). (3) MPI is genuinely a member of that set and co-occurs in the HIF1A-peak and hypoxia-up supplements (R3, partial -- membership confirmed, 'hub centrality' not recomputed). (4) Minor inconsistency flagged: text says SEVEN angiogenesis signatures but the shipped output uses EIGHT (R4, mismatch) -- a numeric inconsistency for the authors/reviewer, below a fabrication threshold. NOT ATTEMPTED (deliberate 80/20 + out of scope): independently regenerating the AngScores from raw TCGA-CRC + tumor-purity (unnecessary; output ships and count is exactly derivable), and all non-SRGA pipelines (HIF ChIP-seq peak calling, hypoxia DEGs, scRNA-seq, NMF subtyping, HIAscore Cox HR=1.5/1.6, immune deconvolution) plus all wet-lab validation -- each a separate RU. GSE37892 is one of nine exploration cohorts; the reproduced SRGA result is computed on TCGA-CRC.
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: 2026-06-14 ⛓ 016a4b4c6c15
✎ 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
- 2026-06-14
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15no human curator yet
- Last updated
- 2026-08-05
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: opusWhat are the transcriptional mechanisms, regulatory networks, and prognostic value of hypoxia-induced angiogenesis (HIA)-related genes in colorectal cancer, and does MPI serve as a key regulator bridging hypoxia and angiogenesis?
- ★ Twelve HIA-related genes were identified that are transcriptionally activated by HIFs and functionally implicated in angiogenesis in CRC. finding
- ★ A prognostic scoring system (HIAscore) built from HIA-related subtype DEGs stratifies CRC patients, with high HIAscore correlating with poor survival, aggressive phenotype, and immunosuppressive tumor microenvironment. resource
- ★ MPI interacts with lactate dehydrogenase A (LDHA). mechanism
- ★ MPI promotes proliferation and angiogenesis of CRC through phosphorylation/activation of the JAK2/STAT3 signaling pathway. mechanism
- ★ Based on HIA-related gene expression, CRC patients were classified into 3 subtypes with distinct tumor hallmark activations. finding
- Spatial analysis revealed sequestration regions between high-HIAscore epithelial cells and T/I/NK cells, hindering immune infiltration. finding
- The SRGA computational method integrated with ChIP-seq and DEG analysis identifies HIA-related genes. method
- ★ MPI is a novel regulator bridging hypoxia-induced angiogenesis and a promising therapeutic vulnerability in CRC. finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| ChIP-seq (HIF1A and HIF2A) | HCT116 colon cancer cell line | hypoxia | HIF1A/HIF2A binding peaks/density at promoters | GSE200203 (public dataset) |
| bulk RNA-seq | CRC cell lines and HUVEC endothelial cell line | hypoxia induced by various stimuli | differential mRNA expression (log2FC) | GSE158632, GSE200204, GSE89831 |
| single-cell RNA-seq | 10 untreated CRC patients | none | cell-type clustering, HIAscore per cell, DEGs | GSE205506 / Seurat |
| spatial transcriptomics | CRC tissue samples | none | spatial localization of HIAscore and deconvolved cell types | Zenodo 10.5281/zenodo.7551712 / SPOTlight, Seurat |
| microarray transcriptome profiling | 11 CRC patient cohorts (TCGA COAD/READ plus GEO) | none | gene expression, clinical/survival association | GEO (GSE37892, GSE39582, etc.); TCGA via UCSC Xena |
| RT-qPCR | HCT116 cells | MPI/HIF-1α shRNA knockdown; CoCl2 hypoxia | relative mRNA expression normalized to β-actin | CFX96 Real-Time PCR System (Bio-Rad); ChamQ SYBR (Vazyme) |
| Co-immunoprecipitation and Western blot | HCT116 cells | MPI knockdown | protein-protein interaction (MPI-LDHA), protein levels / JAK2/STAT3 phosphorylation | BCA Protein Assay Kit; RIPA lysis |
| Tube formation assay | HUVECs pre-treated with HCT116 supernatant | MPI knockdown in HCT116 | total number of tube junctions (angiogenesis) | Matrigel (Corning); Calcein AM |
- – Twelve genes overlapped across the 3 identification methods and were defined as HIA-related genes. 12 genes
- ▲ Patients with high HIAscore showed poor survival, aggressive phenotype, and immunosuppressive microenvironment.
- – MPI co-immunoprecipitates with LDHA, indicating physical interaction.
- ▼ MPI knockdown impaired proliferation and angiogenesis of CRC via the JAK2/STAT3 pathway.
- – CRC patients clustered into 3 HIA-related subtypes (optimal k=3) with distinct clinical and biological features. 3 subtypes
- – Spatial sequestration regions separate high-HIAscore epithelial cells from T/I/NK cells.
- count over 1.9 million new CRC cases (global CRC incidence in 2022)
- count nearly 1 million deaths (global CRC deaths in 2022)
- other 20–45% (proportion of CRC patients with tumor recurrence or metastasis)
- other less than 20% (5-year overall survival rate of advanced CRC)
- count 12 (HIA-related genes identified)
- count 227 (overlapped subtype-related DEGs among three subtypes)
- count 56 (prognosis-related genes used in PCA for HIAscore)
- fold_change log2 fold change > 1, adjusted p < 0.05 (DEG significance threshold for hypoxia vs normoxia)
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 multi-omics computational study integrates ChIP-seq, bulk RNA-seq, scRNA-seq, spatial transcriptomics, and microarray data from multiple public CRC cohorts to identify hypoxia-induced angiogenesis (HIA)-related genes, construct molecular subtypes, and build a PCA-derived prognostic scoring system (HIAscore). Bioinformatic analyses included DESeq2-based DEG analysis, ssGSEA/GSEA enrichment scoring, consensus clustering, and univariate/multivariate Cox regression across multiple independent cohorts. Survival analyses used Kaplan-Meier curves with log-rank tests, supplemented by in vitro and in vivo experiments for mechanistic validation of MPI.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| DESeq2 Wald test | Differential expression analysis of hypoxic vs. normoxic CRC cells (GSE200204 raw counts); subtype-related DEG identification via R/limma | — | not stated |
| Wilcoxon rank-sum test | scRNA-seq DEGs between HIAscore-High and HIAscore-Low epithelial subclusters (Seurat FindMarkers) | 10 CRC patients (GSE205506) | not stated |
| Log-rank test | Kaplan-Meier survival comparison across 3 HIA-related CRC subtypes and HIAscore-High vs. HIAscore-Low patient groups | — | not stated |
| Univariate Cox regression | Selection of prognosis-related genes from 227 subtype-related DEG candidates (GSE_merged cohort) | — | not stated |
| Multivariate Cox regression | Prognostic analysis with clinical covariates (R/forestplot) | — | not stated |
| ssGSEA (single-sample GSEA) | Per-sample hypoxia/angiogenesis signature activity scoring; 28-signature immune cell infiltration estimation via TISIDB signatures | — | na |
| GSEA (gene set enrichment analysis) | Angiogenesis signature scoring per gene within the SRGA method; GO and KEGG functional enrichment analyses (R/clusterProfiler, Metascape) | — | na |
-
Univariate Cox regression with an unadjusted p < 0.05 threshold was used to select 56 prognosis-related genes from 227 subtype-related DEG candidates↳ Could also: LASSO-penalized Cox regression could also be used for prognostic gene selection in this high-dimensional, potentially collinear setting — LASSO Cox simultaneously handles correlated gene expression features, penalizes overfitting via cross-validated shrinkage, and produces a sparse weighted risk score whose coefficients are directly tied to survival outcome — it also avoids the implicit multiple-comparison concern inherent in 227 separate unadjusted univariate tests
-
HIAscore was derived by multiplying gene expression of 56 prognosis-related genes by the sum of their PC1 and PC2 loadings from a PCA↳ Could also: A LASSO/elastic-net Cox risk score, or a simpler mean z-score across the 56 genes, could also summarize the HIA pattern per patient — PCA-based scores capture global variance but lack a direct survival-hazard interpretation; a Cox-penalized score links gene weights explicitly to the outcome being predicted, while a mean z-score is more transparent and easier to implement prospectively in new cohorts
-
Cell-level Wilcoxon rank-sum tests (Seurat FindMarkers) were used to identify DEGs between HIAscore-High and HIAscore-Low epithelial subclusters across 10 patients↳ Could also: A pseudo-bulk approach — aggregating counts per patient per subcluster and then applying DESeq2 or edgeR — could also identify DEGs in scRNA-seq data — Cell-level tests treat individual cells as independent observations, which inflates the effective sample size given the true unit of biological replication is the patient; pseudo-bulk methods aggregate to the patient level and typically produce better-calibrated false-positive rates, particularly when patient numbers are small
-
Consensus clustering (ConsensusClusterPlus) applied to the expression of 12 HIA-related genes was used to define 3 molecular subtypes↳ Could also: Non-negative matrix factorization (NMF) or model-based clustering (e.g., R/mclust) could also be used to identify molecular subtypes from gene expression data — NMF provides a biologically intuitive additive, parts-based decomposition; model-based clustering assigns probabilistic memberships and uses information criteria (BIC) to formally select the number of clusters, offering an alternative statistical basis for the k = 3 choice
-
ssGSEA was used to score hypoxia and angiogenesis pathway activity per sample and to quantify 28 immune cell infiltration signatures↳ Could also: GSVA, AUCell (especially for sparse scRNA-seq data), or per-gene-set mean z-scores could also quantify pathway activity at single-sample resolution — Different scoring methods make different distributional assumptions; GSVA uses a kernel-based rank statistic that can be more sensitive at distributional tails, and AUCell is specifically designed for the high sparsity of scRNA-seq count matrices where ssGSEA rank distributions may be less stable
-
Batch effects across 9 heterogeneous microarray datasets were corrected using ComBat before merging into the GSE_merged discovery cohort↳ Could also: Surrogate variable analysis (SVA), limma's removeBatchEffect, or batch-aware dimensionality reduction (e.g., Harmony for single-cell contexts) could also address technical variation across datasets — ComBat assumes known, discrete batch labels and a parametric empirical Bayes correction model; SVA can additionally estimate latent surrogate variables that capture unmeasured confounders beyond the labeled batch factors, which may be relevant when datasets differ in platform, tissue processing, and patient characteristics simultaneously
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
100 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.
- The consensus molecular subtypes of colorectal cance... 2015 · 3,929 cites
- A colorectal cancer classification system that assoc... 2013 · 771 cites
- Single-cell dissection of transcriptional heterogene... 2011 · 547 cites
- Characterization of the immunophenotypes and antigen... 2015 · 448 cites
- METTL14 suppresses proliferation and metastasis of c... 2020 · 424 cites
- <i>Fusobacterium nucleatum</i> promotes colorectal c... 2019 · 414 cites
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-41204349
Paper: Liu S et al. Multi-omics analyses identify mannose phosphate isomerase (MPI)-centered hypoxia-induced angiogenesis signature in colorectal cancer. J Transl Med 2025. PMID 41204349 · PMCID PMC12595641 · DOI 10.1186/s12967-025-07291-8.
Named code artifact: https://github.com/LucasLiu20200131/SRGA (commit pinned at run time). SRGA = "Signature-Related Gene Analysis", an R package authored by Sicheng Liu (= paper's first author "Liu S"). It is therefore the authors' own analysis tool, and also valid as a standalone third-party tool per the brief (P16).
Reported data availability: "All data ... publicly available ... UCSC Xena, GDC, GEO." Code: "SRGA is available on the Github website (...SRGA/). Any additional information is available upon reasonable request."
What SRGA actually computes (read from R/ source)
correlation_calculation()— Pearson / partial correlation (covariate = tumor purity) across all genes of an expression matrix; relative scoreRS = -log10(p)*sign(cor).fgsea_calculation()— for each selected gene, ranks the other genes by RS and runsfgsea(signatures, ranks, minSize=1, maxSize=5000, nperm=10000);sigValue = -log10(padj)*NES; optional rescale to [-1,1].rank_vis()/col_vis()/net_vis()— rank genes by mean per-signature RRS; count related genes (pval<0.05) per signature; top-gene network.
The repo ships a complete, deterministic worked example (data/exprs.rda,
covariate.rda, Sene.marker.rda; README uses set.seed(1)).
In scope (pipeline-derived, attempted)
| # | Result | Pipeline | Feasibility |
|---|---|---|---|
| R1 | SRGA code artifact runs and is reproducible: documented worked example produces a stable signature-gene ranking + per-signature related-gene counts. | SRGA (shipped data + README example) on «our HPC» | HIGH — bundled data, fixed seed. Primary clear data point. |
In scope but NOT fully reproducible (the hard 20% — documented, not chased)
| # | Reported result | Why not 1:1 |
|---|---|---|
| R2 | "911 unique genes ... angiogenesis-related" from TCGA-CRC (abs score>0.95 & FDR<0.05). | Inputs underspecified: the seven MSigDB angiogenesis signatures are not named; the tumor-purity method is not stated; the TCGA-CRC expression source/normalization is not detailed. fgsea nperm is stochastic. Exact 911 not derivable without guessing → would be chasing the 20%. Reported value pinned from supplement for auditability; reproduction graded partial/not-attempted with reasons. |
Out of scope (wet-lab / external / manual — not attempted)
HIF1A/HIF2A ChIP-seq peak calling (GSE200203), hypoxia DEGs (GSE158632/200204/89831), scRNA-seq (GSE205506), CRC molecular subtyping (NMF/consensus on 9 GEO cohorts), HIAscore Cox models (HR=1.5/1.6), TISIDB/ESTIMATE immune deconvolution, all qPCR/IHC/migration wet-lab validation. These are separate pipelines / bench work, each its own reproduction unit; out of scope for this single-tool RU.
Datasets named (reference)
TCGA COAD/READ (UCSC Xena/GDC); GEO exploration: GSE35896, GSE161158, GSE14333, GSE39582, GSE143985, GSE119409, GSE13294, GSE170999, GSE37892 (the RU's nominal accession — one of nine exploration cohorts); validation GSE17356/17357; ChIP GSE200203; scRNA GSE205506; hypoxia GSE158632/200204/89831.
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 reproduction confirms the paper's headline pipeline number exactly — 911 unique angiogenesis-related genes is the unique-symbol dedup of the shipped Supplementary Data 4, whose schema matches the SRGA code, so no fabrication signal. The main caveats are on our methodology side (the 911 was checked against the authors' shipped output, not re-derived from raw TCGA-CRC, and MPI 'centrality' + downstream Cox/multi-omics claims were left out of scope) plus one minor authors-side contradiction (text 'seven' vs shipped 'eight' angiogenesis signatures). Severity is low and magnitude/direction of the central finding holds, but because the core 'MPI-centered' claim is only confirmed as membership and the deviation is real, this is a solid-with-explainable-deviations (yellow), not a clean 1:1.
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.