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

Identification of Proteins Deregulated by Platinum-Based Chemotherapy as Novel Biomarkers and Therapeutic Targets in Non-Small Cell Lung Cancer.

Front Oncol · 2021
81/100 3/4
Why this verdict

The main results reproduced: recomputed values matched the published ones within tolerance.

Reproduced on the brainbox compute brainarbeit.com
How its reproducibility compares
81/100
Reproducibility score
0.4 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 59% of all assessed papers rank 468 of 1173 scored

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

REPRODUCED (partial). The paper's main pipeline-derived computational result is the Fig3A/3B predictive survival analysis on public GEO GSE42127, and it reproduces well. Design = median-split each of 20 deregulated-protein transcripts into High/Low expression, then WITHIN each stratum compare adjuvant-chemo (ACT, n=49) vs observation (OBS, n=127) overall survival by Kaplan-Meier/log-rank (HR<1 = ACT benefit). Ran on «our HPC» SLURM 2230263 (R 4.5.3/GEOquery 2.78.0/survival 3.8.6). Cohort decodes EXACTLY (n=176, ACT=49/OBS=127); the method reproduces identically; the central biomarker message holds 20/20 (every reported-significant stratum reproduces HR<1, ACT better); 30/40 printed log-rank P-cells are significance-concordant and 12/20 genes reproduce the exact significant stratum at P<0.05, 4 more get the right stratum/direction sub-threshold, and 4 (TMEM205,RPS20,CAPRIN1,RPL35) flip — most plausibly because we picked the most-expressed Illumina probe per gene rather than the paper's Supplemental-Table-4 probe. The two text-stated HRs are directionally consistent (TP53I3 0.36 vs 0.41 within-tol; STMN1 0.27 vs 0.08 same direction). NOT 1:1 on every number but clearly the same analysis and the same conclusion; no fabrication indicated (the paper's TP53I3 CI 0.03-0.59 looks like a print error). NOT attempted: C1 SWATH-MS proteomics (commercial ProteinPilot dependency; PXD024209 public but upstream not reproducible) and C2 druggability ML (repo 404/gone).

💻 Code ↗ 🗄 Data: GSE42127

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.

✎ 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
2026-06-24
Rubric version
not recorded
Assessed by
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 study tests whether cisplatin-deregulated protein networks in non-oncogene driven NSCLC contain novel druggable targets and biomarkers that could predict or improve response to platinum-based chemotherapy.

Core claims
  • Cisplatin exposure induces significant deregulation of protein expression networks in NSCLC cells finding
  • 65 proteins were significantly deregulated (q<0.1) following cisplatin exposure in H460 NSCLC cells finding
  • A machine learning model derived from known druggable/non-druggable protein sequences can predict druggability of the deregulated proteins method
  • Several deregulated proteins (e.g., DPYSL2, ALDH3A1, NUDC, RACK1) show high druggability scores, ligandable structural pockets, and known small-molecule ligands resource
  • Deregulated proteins map to DNA damage response, cell cycle, TGF-beta signaling and apoptosis pathways implicated in platinum resistance mechanism
  • Transcript expression of the deregulated proteins may serve as prognostic biomarkers for survival following adjuvant platinum-based chemotherapy finding
  • ALDH3A1, TP53I3 and FDXR are the top upregulated proteins, while SRXN1, HSP90AA1 and PHGDH are the top downregulated proteins after cisplatin exposure finding
Experimental setups
Assay System Perturbation Readout Platform
SWATH-MS quantitative proteomics H460 NSCLC cell line cisplatin 7.5 µM, 24 h differential protein abundance AB Sciex 5600+ TripleTOF MS with Ekspert NanoLC; ProteinPilot 5.0; Skyline
Western blot NSCLC cells (lysates) cisplatin 5 µM, 12 h protein levels of ALDH3A1 and TP53I3 Odyssey CLx imaging system; ImageJ densitometry
Machine learning druggability prediction in silico protein sequences (FASTA) none druggability classification score (0-1) python scripts, 13 ML classifiers (github.com/muntisa/machine-learning-for-druggable-proteins)
Structural druggability assessment in silico protein structures none ligandable cavities, PDB codes, protein-protein interactome size CanSAR knowledgebase
Ligand/compound database search in silico none known investigational/approved chemical entities targeting proteins BindingDB
GO/KEGG pathway enrichment analysis H460 proteomics dataset (computational) none overrepresented biological processes/pathways ClueGo v2.5.6 in Cytoscape v3.7.2
Canonical pathway/signaling network analysis H460 proteomics dataset (computational) none pathway overrepresentation and activation/inhibition z-scores Ingenuity Pathway Analysis (IPA), QIAGEN
Survival analysis (Kaplan-Meier, log-rank) NSCLC patient tumor tissue, UT Lung SPORE cohort adjuvant platinum-based chemotherapy vs. observation overall survival stratified by transcript expression Illumina gene expression array; GEO dataset GSE42127
Key results
  • 1081 proteins robustly identified by SWATH-MS; 430 upregulated and 586 downregulated after cisplatin exposure
  • 65 differentially regulated proteins reached statistical significance (q<0.1): 26 upregulated, 39 downregulated
  • ALDH3A1, TP53I3 and FDXR were the top three upregulated proteins by log2 fold change and FDR significance
  • SRXN1, HSP90AA1 and PHGDH were the top three downregulated proteins
  • DPYSL2 had the highest druggability score among upregulated proteins 0.999999953
  • NUDC had the highest druggability score among downregulated proteins 0.999999993
  • ITGB1 showed a large CanSAR protein-protein interactome 468 interactions with 307 interactors
  • Survival analysis was performed comparing observation (OBS) versus adjuvant chemotherapy (ACT) patient groups using median-stratified transcript expression
Key statistics
  • pvalue q-value ≤ 0.1 (significance threshold for differentially regulated proteins after cisplatin)
  • count 430 upregulated / 586 downregulated of 1081 total proteins (cisplatin-induced protein changes in H460 cells)
  • count 26 upregulated / 39 downregulated (breakdown of significantly deregulated proteins)
  • other 0.999999953 (DPYSL2 machine-learning druggability score)
  • other 0.999999993 (NUDC machine-learning druggability score)
  • count 127 OBS patients, 49 ACT patients (UT Lung SPORE cohort used for survival analysis (GSE42127))
  • other 7.5 µM (24 h) for proteomics; 5 µM (12 h) for western blot (cisplatin treatment concentrations/durations used)
  • other 18.4% of cancer deaths; NSCLC = 85% of lung cancer cases; 5-year survival 20.5% (epidemiological background on lung cancer burden)

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 used a triplicate, cell-line-based quantitative proteomics design (SWATH-MS) comparing cisplatin-treated versus untreated NSCLC cells, with differential protein expression assessed by empirical Bayes moderated t-statistics (limma) and Benjamini-Hochberg FDR correction (q-value ≤ 0.1). Downstream functional characterization used Fisher's exact test and Z-score prediction (IPA) for pathway analysis and two-sided hypergeometric testing with Bonferroni step-down correction (ClueGo) for GO/KEGG enrichment. A separate clinical cohort analysis stratified patients by median transcript expression and compared overall survival using Kaplan-Meier curves and the log-rank test.

Replicationmixed Sample sizeProteomics: samples prepared 'in triplicate' per condition; survival cohort: 127 patients (OBS) and 49 patients (ACT) from GSE42127; no power calculation described GroupsCisplatin-treated vs untreated NSCLC cells (proteomics); high vs low transcript expression within OBS and ACT groups (survival) Pairingunclear Randomization/blindingnot stated Dispersionunclear Exact p-valuesno Effect sizesyes Confidence intervalsno Multiplicity correctionBenjamini-Hochberg FDR (limma comparison); Bonferroni step-down correction (ClueGo enrichment)
Statistical tests used
Test Applied to n Assumptions
Empirical Bayes moderated t-statistic (limma, R) Differential protein abundance, cisplatin-treated vs untreated H460 cells (SWATH-MS) Triplicate samples per condition not stated
Right-tailed Fisher's exact test IPA canonical pathway over-representation analysis not stated
Two-sided hypergeometric test (enrichment/depletion) ClueGo GO biological process and KEGG pathway enrichment not stated
Log-rank test (with Kaplan-Meier estimation) Overall survival comparison, high vs low transcript expression, in OBS (n=127) and ACT (n=49) cohorts (GSE42127) 127 (OBS), 49 (ACT) not stated
Approaches that could also have been used
  • Differential protein abundance between cisplatin-treated and untreated cells (triplicate samples) was assessed with limma's empirical Bayes moderated t-statistic.
    Could also: A standard two-sample t-test or a non-parametric approach such as the Mann-Whitney U test could also be applied — Moderated t-statistics like limma's are often preferred for small-n proteomics/genomics data because they borrow information across features to stabilize variance estimates, whereas a per-feature t-test or a non-parametric test would not use this shared-information approach but may be more familiar or robust to distributional assumptions with very small n
  • The false discovery rate for the proteomics comparison was controlled using the Benjamini-Hochberg procedure.
    Could also: Other FDR or family-wise error methods, such as Storey's q-value approach or Bonferroni correction, could also be used — Storey's q-value can offer increased power by estimating the proportion of true null hypotheses directly from the data, while Bonferroni offers a more conservative family-wise error control; the choice affects the balance between sensitivity and specificity in flagging significant proteins
  • Pathway over-representation was tested with a right-tailed Fisher's exact test on the significant protein list (IPA), which relies on a fixed significance cutoff to define the input gene/protein set.
    Could also: A rank-based enrichment method such as Gene Set Enrichment Analysis (GSEA) could also be used — GSEA considers the full ranked list of proteins rather than only those passing a significance threshold, which can capture coordinated but sub-threshold changes across a pathway
  • GO/KEGG term enrichment in ClueGo used a two-sided hypergeometric test with Bonferroni step-down correction.
    Could also: A Benjamini-Hochberg FDR correction could also be applied in this context — FDR-based correction is generally less conservative than Bonferroni step-down when testing many overlapping GO/KEGG terms, and could be considered when prioritizing sensitivity to detect enriched terms
  • Survival was compared between patients stratified into high versus low expression groups by the median, using Kaplan-Meier curves and the log-rank test.
    Could also: A Cox proportional hazards regression could also be used, treating expression as a continuous variable or adjusting for covariates — Cox regression provides a hazard ratio with a confidence interval and allows adjustment for potential confounders (e.g., stage, age), whereas median-split log-rank testing offers a simpler, threshold-based comparison without effect-size quantification
  • Densitometric western blot analysis (ImageJ) was described without a stated statistical test for these specific comparisons in the excerpted text.
    Could also: A paired or ratio t-test, or a non-parametric equivalent, could also be applied to densitometry replicates when comparing treated vs untreated lysates — Explicit statistical testing of densitometry data (with an appropriate dispersion measure such as SD or SEM) can help quantify the certainty of observed band-intensity differences across replicates
Software: R / limma R 3.5.2 · Skyline (SWATH-MS quantification) · ProteinPilot (SCIEX) 5.0 · Cytoscape / ClueGo Cytoscape 3.7.2 / ClueGo 2.5.6 · Ingenuity Pathway Analysis (IPA, QIAGEN) · ImageJ

What was reproduced

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

Figures / tables: Fig3Fig3AFig3B
C_N
Reported
176 evaluable NSCLC (127 OBS / 49 ACT)
Reproduced
176 samples; ACT=49, OBS=127
exact
C_method
Reported
median-split High/Low; within-stratum ACT-vs-OBS overall-survival KM/log-rank
Reproduced
reproduced identically (coxph + survdiff per stratum)
exact
C_directional_HRlt1
Reported
ACT-better (HR<1) in indicated stratum for all 20 transcripts
Reproduced
20/20 reported-significant strata reproduce HR<1
exact
C_sig_concordance
Reported
40 per-cell log-rank P (Fig3A/3B High & Low)
Reproduced
30/40 cells significance-concordant; 12/20 genes reproduce the exact significant stratum at P<0.05; 4 partial; 4 flipped (TMEM205,RPS20,CAPRIN1,RPL35)
partial
C_TP53I3_HR
Reported
TP53I3 High HR=0.41 (0.03-0.59) P=0.002
Reproduced
HR=0.361 (0.137-0.952) P=0.031
within tolerance
C_STMN1_HR
Reported
STMN1 Low HR=0.08 (0.01-0.62) P=0.002
Reproduced
HR=0.267 (0.062-1.163) P=0.059 (same direction, sub-threshold)
partial

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

No assessment has been recorded yet.
🤝
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.

241 k
tokens (I/O) · 14.2 M incl. cache
56 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.