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

Prioritized mass spectrometry increases the depth, sensitivity and data completeness of single-cell proteomics.

Nat Methods · 2023
L1 90/100 3/4
Why this verdict

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

Reproduced on the brainbox compute brainarbeit.com
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score -5
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • 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
What did not (or only partly)
  • 🟡A deviation arose in the data or preprocessing
How its reproducibility compares
90/100
Reproducibility score
0.9 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 79% of all assessed papers rank 211 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 1:1. Applied the repo's shipped figGen R pipeline (SlavovLab/pSCoPE @27d5e7d) to its own Zenodo figure-generating data (record 7455405, sha256 8571b1b7...) on «our HPC» compute nodes (R 4.3.3). 9/10 quantitative pipeline-derived claims reproduced exactly or within rounding: Fig1b consistency 18->59%, Fig2a proteins/cell +106% (1431 vs 694) and peptides/run +103% (5549 vs 2729) and MS2-assigned 83.6%, Fig2c/d completeness +171% (peptides) and 93%/+34% (proteins), Fig3b spike-in slope 1.06/R^2 0.97, Fig4 BMDM 1123 proteins x 373 single macrophages. Only divergence: Fig2b median precursor-intensity fold = 2.79 reproduced vs 2.5 reported (partial, same direction; likely rounding/wording). Not attempted: wet-lab MS acquisition + MaxQuant raw search (out of scope, raw spectra not in deposit) and interpretive GO/PSEA biology chunks. No fabrication indicators - every checked value is derivable from shipped data.

💻 Code ↗ 🗄 Data: 10.5281/zenodo.7498141

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
2026-06-30
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-30
no 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: sonnet
Founding hypothesis

The paper tests whether a multi-tiered 'prioritized' precursor-selection strategy (pSCoPE) can simultaneously increase the consistency, sensitivity, depth, and proteome coverage of single-cell proteomics beyond conventional topN shotgun MS, and whether this enables resolving continuous (rather than dichotomous) polarization states in primary macrophages.

Core claims
  • pSCoPE (prioritized precursor selection via MaxQuant.Live) increases sensitivity, data completeness, and proteome coverage more than twofold over shotgun single-cell proteomics finding
  • Prioritization introduces priority levels that define the temporal order of peptide MS2 analysis, favoring high-priority precursors over the topN abundance-based heuristic method
  • Within LPS-treated and untreated macrophages, proteins covary in functional sets (e.g., phagosome maturation, proton transport) similarly across both conditions, linked to phenotypic variability in endocytic activity finding
  • pSCoPE enables quantification of proteolytic products, suggesting a gradient of cathepsin activities within a treatment condition finding
  • pSCoPE quantification shows good accuracy and precision against known spike-in peptide standards across a 16-fold dynamic range finding
  • Prioritization outperforms isobaric match between runs (iMBR) for recovering missing data, and unlike iMBR can be targeted to biologically selected peptides finding
  • pSCoPE is freely available software, widely applicable for analyzing proteins of interest without sacrificing proteome coverage resource
  • Single-cell-derived LPS/untreated fold changes correlate positively with bulk-sample fold changes, supporting measurement accuracy finding
Experimental setups
Assay System Perturbation Readout Platform
Prioritized vs non-prioritized LC-MS/MS benchmarking (MaxQuant.Live real-time control) bulk sample diluted to single-cell levels none (prioritization on/off) peptide/protein identification consistency, data completeness, MS2 analysis rate MaxQuant.Live
Single-cell proteomics (shotgun vs pSCoPE), nPOP sample prep HEK293 and melanoma single cells none (cell type comparison) MS2 spectra assigned to peptide, peptides/proteins quantified per cell, precursor intensity/dynamic range 60-min active gradient LC-MS, 0.5-Th isolation windows
Spike-in peptide quantification accuracy assay single-cell pSCoPE sets (BMDM background) spiked-in peptide standards at known concentrations (5 levels, 16-fold range) measured vs mixing ratio (regression slope, R2) pSCoPE/MS
Single-cell proteomics (SCoPE2 then pSCoPE) murine bone marrow-derived macrophages (BMDMs) LPS treatment (24 h) vs untreated protein abundance, PCA clustering, protein covariation (proton transport, phagosome maturation, IFN signaling) nPOP + MS (SCoPE2/pSCoPE)
Bulk proteomic comparison bulk macrophage samples (M-CSF differentiated) LPS treatment vs untreated protein fold change (LPS/untreated), projected onto single-cell PCA space MS (bulk)
Key results
  • Data completeness for high-priority group of 4,000 peptides increased to 72% with prioritization vs 49% without
  • Fraction of peptides identified in 100% of six runs at 1% FDR rose from 18% (no prioritization) to 59% (prioritization) 228% increased consistency
  • pSCoPE increased fraction of MS2 spectra assigned to a confident peptide sequence to 84%, more than double shotgun over twofold
  • Number of unique peptides per run and quantified proteins per single cell increased with pSCoPE vs shotgun 103% (peptides), 106% (proteins)
  • Median precursor intensity of peptides quantified by pSCoPE was lower than shotgun, indicating wider dynamic range 2.5-fold lower
  • Spike-in measured abundances showed linear dependence on spike-in levels slope=1.06, R2=0.97
  • Single-cell and bulk LPS/untreated fold change estimates correlated positively Spearman ρ=0.91, P=2×10^-11
  • Proton-transport proteins covaried within a treatment condition despite not changing across conditions ρ=0.5, P<10^-12
Key statistics
  • correlation ρ = 0.91, P = 2 × 10−11 (correlation of LPS/untreated protein fold changes between single cells and bulk samples)
  • correlation ρ = 0.5, P < 10−12 (covariation of proton-transport proteins within a treatment condition)
  • fold_change twofold increase in sensitivity, data completeness and proteome coverage (pSCoPE vs shotgun overall performance)
  • fold_change 2.5-fold lower median precursor intensity (pSCoPE vs shotgun median precursor abundance)
  • other regression slope = 1.06, R2 = 0.97 (spike-in peptide quantification accuracy)
  • count 1,123 proteins across 373 single primary macrophage cells (pSCoPE macrophage dataset scale)
  • mean 93% protein-level data completeness (34% gain over shotgun) (pSCoPE data completeness at protein level)
  • count 97% of 4,000 high-priority peptides sent for MS2 analysis (prioritization efficiency for top priority tier)

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 paper is primarily a methods-development study comparing a new mass-spectrometry acquisition strategy (pSCoPE) to standard shotgun proteomics, and secondarily an application of that method to macrophage biology. Results are reported mainly through descriptive summaries (box plots with median/IQR, or mean/median with s.d. error bars) across replicate experiments and single cells, together with Spearman correlations and a linear regression to assess quantitative accuracy of spike-in standards. Multiplicity of correlations/covariation analyses is addressed via q-values, and peptide/protein identifications are filtered at a stated 1% FDR.

Replicationmixed Sample sizeSample sizes are reported as counts of experiments, runs, or single cells per comparison (e.g., n=6, 8, 97, 177, 194, 206, 373), but no a priori power analysis or sample-size justification is described in the excerpted text. GroupsPrioritized (pSCoPE) vs. shotgun/non-prioritized MS acquisition; LPS-treated vs. untreated macrophages; HEK vs. melanoma single cells; spike-in dose levels Pairingunclear Randomization/blindingstated Dispersionmixed Exact p-valuesyes Effect sizesyes Confidence intervalsno Multiplicity correctionq-values (FDR-based) for correlation/covariation analyses; 1% FDR threshold (target-decoy) for peptide/protein identification
Statistical tests used
Test Applied to n Assumptions
Simple linear regression (slope and R^2) Fig. 3b, agreement between spike-in mixing ratios and measured reporter-ion intensities eight prioritized single-cell experiments (pSCoPE sets), five spike-in levels not stated
Spearman rank correlation Fig. 4c, correlation between LPS-treated/untreated protein fold changes estimated from single cells vs. bulk samples 28 proteins not stated
Spearman rank correlation Fig. 4d, covariation among proton-transport (V-ATPase) proteins within a treatment condition 177 single cells not stated
Descriptive box-plot summaries (median, 25th/75th percentile, 1.5xIQR whiskers) Figs. 1b, 1c, 2a comparing prioritized vs. shotgun/non-prioritized acquisition six experiments (Fig. 1b/c); eight experiments or 97 single cells (Fig. 2a) na
Principal component analysis (PCA) with protein set enrichment analysis (PSEA) Figs. 2e and 4a/4b, clustering of single cells by cell type or treatment and identification of enriched protein sets along PCs 97, 194/206, and 373 single cells depending on panel not stated
Approaches that could also have been used
  • Comparisons between prioritized and shotgun/non-prioritized acquisition (Figs. 1b, 1c, 2a) are shown as box-plot summaries without an accompanying inferential test statistic or p-value.
    Could also: A paired or unpaired nonparametric test (e.g., Wilcoxon signed-rank or Mann-Whitney U) or a mixed-effects model accounting for experiment-level clustering — This would let readers see, alongside the descriptive box plots, a formal estimate of how unlikely the observed difference in medians would be under a null of no difference, and a mixed-effects approach would also account for repeated measurements coming from the same experiments.
  • Dispersion is reported as s.d. in some figures (Figs. 2b, 3b) and as IQR via box-plot whiskers in others (Figs. 1b, 1c, 2a).
    Could also: A single consistent dispersion measure throughout, such as a 95% confidence interval or IQR everywhere — A uniform reporting convention (e.g., CIs) directly conveys the precision of an estimate and its likely range for replication, and can make cross-figure comparisons more direct for readers.
  • Multiple correlation/covariation analyses (Fig. 4c, 4d, and related supplementary figures) are each associated with a q value, described as an FDR-type control.
    Could also: An explicitly named and cited multiple-testing procedure, such as Benjamini-Hochberg FDR or Bonferroni correction, applied across the full family of correlation tests performed — Naming the specific method and the exact family of tests it was applied to would let readers reproduce the multiplicity adjustment and assess its scope.
  • Quantitative accuracy of pSCoPE was assessed via a linear regression of measured vs. spike-in abundance (slope = 1.06, R^2 = 0.97), without reported uncertainty on the slope estimate.
    Could also: Reporting a confidence interval around the regression slope/intercept, or using a Deming/orthogonal regression given that both axes carry measurement error — A CI on the slope would quantify how precisely the 1:1 relationship is estimated, and an errors-in-variables regression is often preferred when both the reference and the measured values are subject to experimental noise.
  • Cluster separation by cell type or treatment condition (Figs. 2e, 4a) is shown via PCA visualization and projected bulk samples, without a formal statistical test of group separation.
    Could also: A permutation-based multivariate test such as PERMANOVA, or a classification-based cross-validation accuracy, applied to the PCA or full proteomic profile — This would provide a quantitative, testable statement of how distinguishable the groups are in the high-dimensional protein space, complementing the visual PCA clustering.
  • Associations between single-cell and bulk fold-change estimates, and between covarying proteins, are summarized using Spearman correlation only.
    Could also: Reporting Pearson correlation alongside Spearman (when linearity is of interest), or a robust/percentage-bend correlation for additional sensitivity to outliers — Reporting both rank-based and linear correlation coefficients can clarify whether the relationship is monotonic versus linear, which is useful when readers want to model the reported fold-change relationship directly.
Software: MaxQuant.Live

What was reproduced

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

Scope — pmid-37012480 (pSCoPE, Nat Methods 2023)

Title: Prioritized mass spectrometry increases the depth, sensitivity and data completeness of single-cell proteomics. PMID 37012480 · PMCID PMC10172113 · DOI 10.1038/s41592-023-01830-1 Code: https://github.com/SlavovLab/pSCoPE (commit 27d5e7d9ae581c0d54cc851826151efc17d0094d, 2024-12-07) Data (as cited): zenodo 10.5281/zenodo.7498141 (brief) / badge 10.5281/zenodo.7498171 (README) Data ACTUALLY downloaded by the code: https://zenodo.org/record/7455405/files/pSCoPE_figureGeneratingData.zip (1,086,912,266 bytes; this is the URL hard-coded in codeBlocks/pSCoPE_figGen_Zenodo_v1.rmd). The three Zenodo IDs are versions of the same concept-record; the figure-generating data is record 7455405.

How the paper's computational results are produced

The repository's figure-generation notebook codeBlocks/pSCoPE_figGen_Zenodo_v1.rmd (R, ~3,800 lines) downloads pSCoPE_figureGeneratingData.zip from Zenodo, unzips it to datIn/, and regenerates every main + extended-data figure from already-search-engine-processed MaxQuant output tables (evidence.txt, msms.txt, msmsScans.txt, peptide/protein matrices) plus MaxQuant.Live instrument log files. The heavy wet-lab + raw-MS search steps (MaxQuant.Live acquisition on a Thermo Orbitrap, MaxQuant database search) are NOT in this repo — only the downstream figure pipeline is.

In scope (pipeline-derived, reproducible here)

Running the R figGen pipeline on the shipped datIn/ tables regenerates these quantitative results:

ID Result Fig Pipeline
C1 Proteins per run: prioritized vs not-prioritized (median) Fig 1b figGen Technical Figure 1b chunk on Fig1/MQL_contrast_evidence.txt
C2 High-priority-peptide data completeness / success-rate distribution Fig 1b/1c figGen Fig1b/1c chunks
C3 Quantified proteins per single cell: pSCoPE vs shotgun (+106%) Fig 2a figGen Fig2a on Fig2/Fig2_a_Coverage/*
C4 Unique peptides per run (+103%) / MS² assignment fraction 84% Fig 2a figGen Fig2a
C5 Median precursor intensity ~2.5-fold lower (pSCoPE vs shotgun) Fig 2b figGen Fig2b
C6 Data completeness gain for challenging peptides (+171%) Fig 2c figGen Fig 2c
C7 Protein-level data completeness 93% (vs shotgun), +34% Fig 2d figGen Fig 2d
C8 Spike-in quantification: slope 1.06, R²=0.97, 16-fold range Fig 3b figGen Fig 3b
C9 1,123 proteins quantified across 373 single macrophages Fig 4a figGen BMDM data-import chunk
C10 Twofold (>2×) overall gain in sensitivity/completeness/coverage Abstract aggregate of C3–C7

Out of scope (not attempted, why)

  • Wet-lab / instrument acquisition (MaxQuant.Live prioritized acquisition on Orbitrap; sample prep; nPOP). Hardware + proprietary software — not reproducible.
  • MaxQuant database search of raw .raw files into the evidence/msms tables. The raw files (~hundreds of GB) live on MassIVE, not in the figGen deposit; the paper's figGen pipeline starts from the searched tables. We reproduce from those tables (P16: applying the shipped pipeline to the shipped data is valid).
  • Gene-ontology / PSEA biological interpretation chunks (Fig 4–6 biology) are attempted opportunistically but down-ranked — they need Bioconductor GO DBs and are interpretive rather than a single pinnable number.

Reproduction strategy

  1. Build R env (conda) on «our HPC» front1; data already on «infra».
  2. Patch the notebook to read datIn/ locally (skip the Zenodo download chunk, already fetched) and to PRINT/serialize the numeric values feeding each panel.
  3. Run as a SLURM compute job; pull back the small numeric outputs (CSV/JSON).
  4. Compare extracted numbers to the paper claims above (claims.tsv / agreement.json).
Figures / tables: Fig 1bFig 2aFig 2bFig 2cFig 2dFig 3b
C1b
Reported
18%->59% (+228%) consistency
Reproduced
18.2%->59.2% (+225%)
exact
C3
Reported
+106% proteins/cell
Reproduced
1431 vs 694 (+106.2%)
exact
C4
Reported
+103% peptides/run
Reproduced
5549 vs 2729 (+103.3%)
exact
C4b
Reported
84% MS2 assigned
Reproduced
83.6%
within tolerance
C5
Reported
2.5-fold lower precursor intensity
Reproduced
2.79-fold lower
partial
C6
Reported
+171% peptide completeness (challenging)
Reproduced
77.3 vs 28.6% (+170.7%)
exact
C7
Reported
93% protein completeness (+34%)
Reproduced
93.4% (+33.8%)
exact
C8
Reported
spike-in slope 1.06, R^2 0.97
Reproduced
slope 1.06, R^2 0.97
exact
C9
Reported
1123 proteins x 373 single macrophages
Reproduced
1123 x 373
exact
C1
Reported
Fig1b proteins/run prioritized>not (boxplot)
Reproduced
1734 vs 1691.5 (priori higher)
partial
C10
Reported
over twofold overall gain
Reproduced
supported by C3/C4/C6/C7
exact

Assessments & scoring basis

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

🤖 AI curator · claude (ai-curator room) · v1.0 L1 90/100

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.

🟢1. Data identity
🟢2. Endpoint comparability
🟡3. Location of the main deviation
🟢4. Cause of the deviation
🟢5. Derivability / plausibility
🟢6. Severity of the deviation
🟢7. Core claim
🟢8. Severity of the miss (overall human judgment)
Scoring basis — itemised

Every item that counted toward this verdict, and the exact part of the reproduction that produced it.

Supporting (toward a concern)
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Concordant (toward reproduced)
All content-critical questions reproduced
-4 pts
From: Q7 · Core claim 🟢
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score -5

This is a near-textbook 1:1 reproduction: the reproduction applied the authors' own shipped figGen R pipeline to their own deposited Zenodo data and recovered 9/10 quantitative figure claims exactly or within rounding, including the central over-twofold depth/completeness gains (C3 +106%, C4 +103%, C6 +171%, C7 93%/+34%) and exact spike-in regression. The only deviation, C5 precursor intensity (2.79 vs 2.5-fold), is a secondary metric, same direction, and most plausibly a rounding/wording difference on our methodology side — not an authors' defect. No fabrication or non-derivability indicators; every value is reproducible from the shared data.

🤝
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.