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

poreCov-An Easy to Use, Fast, and Robust Workflow for SARS-CoV-2 Genome Reconstruction via Nanopore Sequencing.

Front Genet · 2021
L1 100/100 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)
Content-critical question only partially held
+2 pts
From: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q4 · Cause of the deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q6 · Severity of the deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q2 · Endpoint comparability 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +8
✓ What held up
  • Nothing in this column.
What did not (or only partly)
  • 🟡Could not use the authors’ exact input data
  • 🟡Reported values were only indirectly comparable
  • 🟡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
How its reproducibility compares
100/100
Reproducibility score
1.5 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 95% of all assessed papers rank 1 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. Tool/workflow paper reproduced via P16: ran the authors' own poreCov 1.1.0 (Nextflow 22.10.8 + apptainer, profile test_fastq,local,singularity) on its shipped OSF test data on «our HPC» («job», 16 cores, 10m02s, exit 0). All four in-scope pipeline claims reproduced exactly: C1 consensus is BYTE-FOR-BYTE identical to the author-shipped expected medaka consensus (29903 bp, 189 N, identity 1.00000, 0 diffs); C2 president QC PASS (identity 0.993>=0.90, N 0.63%<=5%); C3 pangolin lineage B.1.177; C4 nextclade clade 20E (EU1). Determinism comes from poreCov 1.1.0 pinning pangolin 3.1.17/2022-01-05 and nextclade 1.10.1 in its containers (--update NOT used). The dataset (OSF kf54a fastq + 87bc9 truth) is open, complete, and delivers what it promises as a CI/validation set. NOT attempted (out of scope): Table 1 (external GISAID metadata, not pipeline-derived), Table 2 (hardware-dependent runtimes), and Fig 2 per-sample/negative-control values (the 4 clinical samples are not publicly deposited; no ENA/SRA accession). All grades are provisional and human-checkable.

💻 Code ↗ 🗄 Data: 10.5281/zenodo.4534746

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.

  1. v1 current initial assessment Score 50
    assessed: 2026-06-18 ⛓ ca82027949f7
✎ 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
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19
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 addresses whether a highly parallel, containerized Nextflow workflow (poreCov) can make accurate, reproducible SARS-CoV-2 genome reconstruction and lineage determination from nanopore (ARTIC protocol) sequencing data accessible to labs lacking dedicated bioinformatics expertise, while remaining fast and robust at high sample throughput.

Core claims
  • poreCov is an easy-to-use, fast, and robust Nextflow-based workflow for reference-based SARS-CoV-2 genome reconstruction and lineage determination from nanopore sequencing data resource
  • poreCov wraps all required tools in version-controlled Docker/Singularity containers, requiring only Nextflow plus a container engine as dependencies method
  • poreCov flexibly accepts fast5, fastq, or fasta inputs and automatically adjusts workflow steps accordingly method
  • poreCov generates a comprehensive HTML/XLSX/TSV summary report combining QC, contamination, lineage, and coverage information for rapid interpretation resource
  • Sample-barcode bleeding can be identified by comparing SNP proportion patterns of negative controls against barcoded samples in BAM files finding
  • Longer 1,200 bp amplicon primer sets reduce amplicon dropouts and increase throughput relative to standard 400 bp ARTIC V3 amplicons finding
  • poreCov has been validated at scale across multiple institutions, processing over 2,000 real SARS-CoV-2 clinical samples finding
  • Requiring barcodes on both read ends mitigates but does not eliminate sample-barcode bleeding, at the cost of more unclassified reads finding
Experimental setups
Assay System Perturbation Readout Platform
GISAID metadata survey 1,185,291 publicly deposited SARS-CoV-2 genome entries none proportion of entries by sequencing technology used GISAID
Workflow runtime benchmarking poreCov pipeline (Nextflow) run on 10 fastq read files varying hardware configuration (threads/RAM) and cores flag time to final report Docker/Singularity on local servers and Google Life Sciences cloud
Real-world clinical sample sequencing and genome reconstruction SARS-CoV-2 PCR-positive patient samples from Jena University Hospital, University Hospital Regensburg, and Germany's Public Health Institute none number of samples/genomes successfully processed MinION/GridION nanopore sequencers
Taxonomic read classification nanopore sequencing reads from patient samples none proportion of human, viral (SARS-CoV-2), and unclassified reads Kraken2 against combined GRCH38.p13 + GISAID SARS-CoV-2 database
Read quality control fastq (basecalled) or fast5 (raw) nanopore reads none read length distribution, quality scores NanoPlot (fastq) / pycoQC (fast5)
Reference-based genome reconstruction and variant calling length-filtered amplicon nanopore reads mapped to Wuhan reference (NC_045512.2) none consensus genome sequence, masked low-coverage regions minimap2, SAMtools/BCFtools, medaka, Longshot, nanopolish (ARTIC pipeline)
Lineage and clade assignment reconstructed/polished SARS-CoV-2 consensus genomes none lineage, clade, mutations, deletions, genome quality metrics pangolin, nextstrain, president
Sample-barcode bleeding assessment negative controls vs. barcoded samples (BAM alignment files) overloaded DNA molecule input during library prep SNP base-proportion patterns per position IGV / Unipro UGENE visualization of BAM files
Key results
  • Illumina was the dominant sequencing technology for SARS-CoV-2 surveillance, followed by nanopore Illumina 77.61% (n=913,228) vs Nanopore 17.21% (n=202,503) of 1,185,291 GISAID entries
  • poreCov runtime decreased substantially with more parallel hardware resources 1 h 23 min 52 s (4 threads) to 7 min 47 s (256 threads), ~10.8-fold reduction
  • poreCov was used in production across multiple institutions on large real-world sample sets >600 samples (Jena), >400 samples (Regensburg), 1,012 samples (Germany's Public Health Institute, Feb 23–May 5, 2021)
  • A negative control showed reads classified as SARS-CoV-2, indicating possible contamination or barcode bleeding 98 reads
  • Genomes are flagged QC pass based on identity and ambiguous base thresholds ≥90% identity to Wuhan reference and ≤5% ambiguous nucleotides (Ns)
  • Low-coverage genome regions are masked during reconstruction coverage ≤20x masked
  • Typical samples yield high proportions of viral reads, while lower proportions correlate with reconstruction failure or low quality typical ≥90% viral reads
Key statistics
  • count 1,185,291 GISAID entries screened (April 2021) (sequencing technology survey)
  • other Illumina 913,228 (77.61%); Nanopore 202,503 (17.21%); SMRT 30,447 (2.59%); Ion Torrent 12,895 (1.10%); DNBSEQ 12,658 (1.08%); Sanger 4,898 (0.42%) (proportion of sequencing technologies used for SARS-CoV-2 surveillance)
  • other runtime from 1 h 23 min 52 s to 7 min 47 s depending on hardware (4 to 256 threads) (poreCov time-to-results benchmarking, 10 fastq files)
  • count 98 reads classified as SARS-CoV-2 in a negative control (example report showing possible contamination/barcode bleeding)
  • count >600 samples analyzed at Jena University Hospital (real-world validation of poreCov)
  • count >400 samples analyzed at University Hospital Regensburg (real-world validation of poreCov)
  • count 1,012 samples analyzed at Germany's Public Health Institute between Feb 23 and May 5, 2021 (real-world validation of poreCov)
  • other QC pass threshold: ≥90% identity to NC_045512.2 and ≤5% ambiguous nucleotides (genome quality control criteria used in poreCov report)

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.

This is a software/methods paper presenting poreCov, a Nextflow-based workflow for SARS-CoV-2 genome reconstruction from nanopore sequencing data; no formal inferential statistical tests are reported. Evaluation is descriptive: runtime benchmarks across five hardware configurations are presented as single-measurement durations (Table 2), and sequencing technology proportions from GISAID are reported as counts and percentages (Table 1). Pipeline validation is conveyed through deployment scale (>600 samples at Jena University Hospital, >400 at University Hospital Regensburg, and 1,012 samples at the Robert Koch Institute) without comparative statistical analysis.

Replicationunclear Sample sizeDeployment counts reported as lower-bound estimates: >600 samples (Jena), >400 samples (Regensburg), 1,012 samples (RKI, Feb–May 2021); runtime benchmarks based on 10 fastq input files per hardware configuration with no stated replication GroupsFive hardware/software configurations compared by wall-clock runtime; no control vs. treatment comparison Pairingna Randomization/blindingna Dispersionnone
Approaches that could also have been used
  • Runtime benchmarks in Table 2 are reported as single wall-clock measurements per hardware configuration
    Could also: Repeated benchmark runs (e.g., n ≥ 3 per configuration) summarized with median and interquartile range or mean ± SD could also characterize throughput — A single timing measurement captures one realization of a stochastic process influenced by I/O scheduling, background load, and container caching; repeated measurements would quantify run-to-run variability and allow comparison intervals rather than point estimates
  • Pipeline performance is validated by reporting the number of samples processed at three sites without a formal comparison to existing tools (e.g., nf-core/viralrecon, ncov2019-artic-nf)
    Could also: A head-to-head benchmarking study on a shared reference dataset with known variant calls could also characterize relative accuracy, sensitivity, and specificity — Deployment counts demonstrate feasibility and scale but do not quantify concordance with a ground truth; a comparative benchmark on, for example, simulated reads or WHO reference panel samples would allow readers to situate poreCov's accuracy relative to other workflows
  • QC pass thresholds are fixed at ≥90% nucleotide identity to the Wuhan reference and ≤5% ambiguous bases (Ns) without derivation
    Could also: A sensitivity analysis varying these thresholds across the >1,000 processed samples, or a receiver-operating characteristic approach against an orthogonal quality metric, could also characterize how threshold choice affects the pass/fail rate — Reporting the empirical distribution of percent identity and N-content across the full sample set would allow readers to understand how conservative or permissive the chosen thresholds are relative to the data actually observed
  • Proportions of sequencing technologies in Table 1 are reported as raw counts and percentages from a GISAID snapshot (April 2021)
    Could also: Confidence intervals for each proportion (e.g., Wilson score intervals) could also be reported alongside the point estimates — With n > 1 million entries, the proportions are estimated with high precision, but explicitly reporting uncertainty would be standard practice for descriptive epidemiological proportions and makes the precision transparent
  • The negative control contamination example (98 reads classified as SARS-CoV-2) is presented qualitatively via visual inspection of SNP proportion patterns in IGV
    Could also: A formal contamination threshold (e.g., reads-above-background defined by a Poisson or binomial model) could also be derived from multiple negative control runs — A quantitative threshold would give users a reproducible criterion for flagging contamination rather than relying on visual inspection, and would allow the sensitivity of the negative-control check to be characterized across different sequencing runs
  • Sample sizes at the three deployment sites are reported as lower-bound estimates (>600, >400, 1,012) without breakdown by variant, primer version, or sequencing device
    Could also: Stratified reporting by ARTIC primer version (V3 vs. 1,200 bp), device type (MinION vs. GridION), or time period could also characterize pipeline performance across subgroups — Pooled deployment counts demonstrate broad applicability but do not reveal whether performance differed across protocol variations; stratified summaries would be informative for labs choosing between primer sets or devices
Software: Nextflow · Kraken2 · minimap2 · medaka · Longshot · nanopolish · SAMtools/BCFtools · NanoPlot · pycoQC · pangolin · nextstrain/nextclade · president · Krona · CoVarPlot

What was reproduced

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

Scope — pmid-34394197 (poreCov)

Paper: Brandt et al. 2021, poreCov—An Easy to Use, Fast, and Robust Workflow for SARS-CoV-2 Genome Reconstruction via Nanopore Sequencing, Front Genet, DOI 10.3389/fgene.2021.711437.

What poreCov is: a Nextflow workflow (DSL2) that wraps existing tools to go from Nanopore reads → SARS-CoV-2 consensus genome → QC + lineage/clade. Wrapped tools (poreCov 1.1.0 pinned containers):

  • ARTIC pipeline artic minion --medaka (genome reconstruction) — nanozoo/artic:1.3.0-dev
  • minimap2 (mapping, inside ARTIC), medaka (consensus polishing), Longshot/medaka variants
  • president rkibioinf/president:v0.6.3 (identity-to-reference + N% QC)
  • pangolin nanozoo/pangolin:3.1.17--2022-01-05 (lineage)
  • nextclade nanozoo/nextclade:1.10.1--2022-01-18 (clade + mutations)
  • kraken2 (read decontamination, OPTIONAL, OFF by default), NanoPlot/pycoQC, Krona, CoVarPlot

This is a tool/workflow paper → P16 reproduction: run the authors' own tool on its own shipped test data and verify the described pipeline-derived behaviour.

Reported results and their reproduction status

# Reported result Location In scope? Pipeline
Table 1 GISAID sequencing-tech distribution (Illumina 77.61% n=913,228; nanopore 17.21% n=202,503 of 1,185,291) Table 1 OUT — external GISAID metadata snapshot, not pipeline-derived
Table 2 Runtime benchmark (4thr/32GB 1h23m; 24thr/64GB 23m; 256thr/1TB 7m47s; GCloud 19m) Table 2 OUT (1:1) — hardware-dependent; we report our wall-clock as a non-comparable sanity datum only nextflow
Fig 2 Sample report for 4 clinical samples + 1 neg control (neg ctrl: 98 reads classified SARS-CoV-2 by kraken2); V1200 primers Fig 2 OUT — clinical samples NOT publicly deposited (no ENA/SRA); screenshot, no per-sample numbers in text poreCov full
QC criteria "QC pass" = ≥90% identity to Wuhan (NC_045512.2) AND ≤5% Ns; coverage masked at ≤20× Methods IN president
Workflow behaviour poreCov reconstructs a QC-passing SARS-CoV-2 consensus + assigns lineage/clade from Nanopore reads Throughout IN (core claim) full poreCov

In-scope reproduction (what we actually run)

Run poreCov 1.1.0 (the version in the documented test command) end-to-end on its own shipped test data via the test_fastq profile:

  • Test reads: https://osf.io/kf54a/downloadSARSCoV2.fastq.gz (183,965 Nanopore reads, sample 2020_12_10_ukj_covid barcode13, V1200/Midnight 1200 bp amplicons).
  • Built-in ground-truth consensus shipped by the authors: https://osf.io/87bc9/downloadSARSCoV2.fasta, header >barcode13/ARTIC/medaka MN908947.3 (~29.9 kb) — this is poreCov's OWN medaka consensus for the same sample, so the fastq→consensus step has an author-provided expected output to compare against.

Reproduction claims:

  • C1 (consensus reconstruction): reads → artic minion --medaka consensus; compare to shipped SARSCoV2.fasta (author ground truth). Metric: length + base identity.
  • C2 (QC / president): consensus identity to Wuhan NC_045512.2 ≥ 90% and Ns ≤ 5% → expect QC pass (the criterion the paper defines).
  • C3 (lineage / pangolin 3.1.17, data 2022-01-05): lineage call for the consensus (deterministic at pinned container data version).
  • C4 (clade / nextclade 1.10.1): clade + mutation count.

Determinism note: pangolin/nextclade calls drift with database version; poreCov 1.1.0 pins these in its containers (pangolin data 2022-01-05, nextclade 1.10.1), so the calls are reproducible at the paper-era pin. We use -r 1.1.0 containers, NOT --update.

Out of scope (not attempted) and why

  • Table 1 GISAID stats: external metadata, not produced by the pipeline.
  • Table 2 timings: hardware-specific; no faithful 1:1 possible on different hardware.
  • Fig 2 per-sample/neg-control numbers: the clinical samples are not publicly deposited (Data Availability points only to the GitHub repo; no
C1-consensus
Reported
poreCov reconstructs the shipped expected medaka consensus (>barcode13/ARTIC/medaka MN908947.3, ~29.9 kb) from Nanopore reads
Reproduced
29903 bp, 189 N; identity to shipped truth = 1.00000 (0 base differences, incl. identical N positions)
exact
C2-qc
Reported
QC pass = >=90% identity to Wuhan NC_045512.2 AND <=5% Ns (president)
Reproduced
identity 0.993, N 189/29903=0.63%, qc_all_valid=True -> QC PASS=YES
exact
C3-lineage
Reported
pangolin lineage at the pinned 3.1.17 / pangoLEARN 2022-01-05
Reproduced
B.1.177 (pangolin 3.1.17, pangoLEARN 2022-01-05, status passed_qc)
exact
C4-clade
Reported
nextclade clade + mutations at the pinned 1.10.1
Reproduced
20E (EU1), QC good, 20 nt / 9 aa substitutions; consistent with B.1.177
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 100/100

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)
Content-critical question only partially held
+2 pts
From: Q5 · Derivability / plausibility 🟡
Content-critical question only partially held
+2 pts
From: Q7 · Core claim 🟡
Content-critical question only partially held
+2 pts
From: Q8 · Severity of the miss (overall human judgment) 🟡
Minor / cosmetic deviation
+1 pts
From: Q3 · Location of the main deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q4 · Cause of the deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q6 · Severity of the deviation 🟡
Minor / cosmetic deviation
+1 pts
From: Q2 · Endpoint comparability 🟡
Concordant (toward reproduced)
Code + data deposited & functional
-2 pts
From: Data & code availability Available & functional
Total score +8

This is a tool/workflow paper (poreCov) reproduced legitimately by re-running the authors' own pinned pipeline (pangolin 3.1.17, nextclade 1.10.1) on its shipped OSF test data, which makes the capability claims deterministic and derivable. The gaps are on the data-availability/our-scope side, not the authors' side: the paper's clinical samples behind Table 1 and Fig 2 were never deposited, so those exact values are uncheckable and were placed out of scope. Severity is unresolved rather than large — the «our HPC» run («job») was still in progress with all claims PENDING and agreement.json not-run-yet. Net: a solid, explainable, but unfinished reproduction with no fabrication signal → yellow across the board.

🤝
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.

126.9 k
tokens (I/O) · 8.6 M incl. cache
12 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.