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

Telomere-to-telomere reference genome for Panax ginseng highlights the evolution of saponin biosynthesis.

Hortic Res · 2024
L1 92/100 3/4
Why this verdict

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

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • 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
  • 🟡A deviation was attributed to the published material
How its reproducibility compares
92/100
Reproducibility score
1.0 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 83% of all assessed papers rank 179 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, strong). All 7 in-scope pipeline claims reproduced on «our HPC» from the published T2T assembly FASTA (figshare 25477741) via the cited tool T2Tools (centel + gapcount) and seqkit. 6/7 EXACT: chromosomes=24, assembly size=3.455 Gb, contig N50=147.37 Mb, GC=34.71%, gaps=0 (T2Tools gapcount), centromeres=24 (T2Tools centel = exactly 24 regions, 1/chr). C1 telomeres within-tol: T2Tools centel calls 45/48, and a tool-independent terminal motif scan confirms all 48/48 chromosome ends carry telomeric arrays, corroborating the paper's reported 48 -- the 3-end gap is a pure tool-sensitivity difference (the paper used quarTeT, which tolerates the degraded 5' arrays on Chr01/Chr05/Chr23 that T2Tools' strict TRF-record criterion rejects), not a data discrepancy or fabrication. The one real compute obstacle -- whole-chromosome TRF on Chr04 (the densest chr) exceeding the 12h cluster wall -- was solved cleanly with T2Tools' own windowed-TRF approach (2 Mb windows, identical params, offset-merged; finished in ~5 min), with the telomere/centromere counts proven window-safe. NOT attempted (out of scope, by 80/20): de novo T2T assembly (hifiasm+Hi-C+gap-filling), genome annotation (77,266 genes; LAI), saponin/DDS evolutionary analyses. C8 BUSCO (99.3%) was an optional stretch outside the core pipeline and was blocked by shared-env storage. Conclusion: the deposited assembly delivers every reported headline metric, and its telomere/centromere structure is faithfully reproduced with the cited tool.

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 92
    assessed: 2026-06-21 ⛓ 82bbf5ebacdf
✎ 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
2026-06-21
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-21
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 study seeks to resolve the complex allotetraploid genetic background of Panax ginseng by assembling a telomere-to-telomere reference genome and using it to determine how the two subgenomes evolved (gene loss, expression bias, divergence time) and how this evolution shaped ginsenoside (triterpene saponin) biosynthesis.

Core claims
  • A telomere-to-telomere reference genome of P. ginseng was assembled (3.45 Gb, 24 chromosomes, 77266 protein-coding genes) resource
  • The genome separates into subgenome A and subgenome B, which show asymmetric gene loss with subgenome B having a general expression advantage despite fewer genes finding
  • Subgenomes A and B diverged approximately 6.07 million years ago, with subgenome B most closely related to Panax vietnamensis var. fuscidiscus finding
  • Gene families associated with ginsenoside biosynthesis (sesquiterpene/triterpene biosynthesis, cytochrome P450) are expanded in both ginseng subgenomes finding
  • Tandem duplications and proximal duplications play crucial roles in ginsenoside biosynthesis gene evolution finding
  • Functional genes located in colinear regions between subgenomes show divergent functions, indicating unbalanced evolution of the saponin biosynthesis pathway finding
  • The ginseng genome contains a special telomeric repeat sequence (AAATTTT)n on chromosome 1, in addition to the conventional (AAACCCT)n repeat finding
  • Metabolic gene clusters (MGCs) for saponin biosynthesis, previously reported in other medicinal plants such as Salvia miltiorrhiza, had not yet been detected in ginseng finding
Experimental setups
Assay System Perturbation Readout Platform
Whole-genome sequencing and de novo assembly Panax ginseng (whole plant/genome) none genome sequence, contig N50, gap filling PacBio HiFi (~140G), Oxford Nanopore (~250G), Illumina (~140G), Hi-C (~700G)
Genome annotation (ab initio + homology + RNA-seq based) Panax ginseng genome none number of protein-coding genes, annotation completeness
RNA-seq read mapping Panax ginseng none mapping rate to reference genome
BUSCO completeness assessment Panax ginseng genome/gene set none % complete single-copy orthologs BUSCO
Telomere and centromere prediction Panax ginseng chromosomes none number/location of telomeres and centromeres quarTeT; T2Tools
K-mer based clustering and phylogenomic tree construction Panax ginseng and Araliaceae species (E. senticosus, A. elata, P. stipuleanatus, P. vietnamensis var. fuscidiscus, P. notoginseng, P. quinquefolius, P. japonicus, V. vinifera, L. sativa, D. carota) none subgenome assignment, species divergence times
Comparative genomics (gene family expansion/contraction, KEGG enrichment, Ks distribution, collinearity/synteny analysis) Ginseng subgenome A vs subgenome B, and vs other Panax/Araliaceae species none expanded/contracted gene family counts, Ks peaks (WGD timing), structural variations (inversions, translocations, duplications)
Pfam domain comparison and Fisher's exact test Ginseng subgenome A vs subgenome B (compared also to P. stipuleanatus homologs) none subgenome-specific Pfam domain counts, retained homologous gene counts, statistical significance
Key results
  • Assembled genome of 3.45 Gb across 24 contigs with contig N50 of 147.37 Mb and 77266 protein-coding genes, an improvement over ginseng v1.0 N50 147.37 Mb vs 19.75 Mb (v1.0)
  • High assembly quality: QV 41.86, K-mer completeness 97.4%, genome BUSCO 99.3%, RNA-seq mapping rate 94-98%, 48 telomeres and 24 centromeres identified BUSCO 99.3%
  • Subgenome A (1.94 Gb, 40550 genes) is larger with more genes than subgenome B (1.51 Gb, 36716 genes), but subgenome B shows a general expression advantage
  • Subgenomes A and B diverged approximately 6.07 million years ago; subgenome B is closest to P. vietnamensis var. fuscidiscus (diverged ~4.98 Mya) 6.07 MYA / 4.98 MYA
  • Subgenome A had 3380 expanded and 5358 contracted gene families; subgenome B had 2344 expanded and 5971 contracted gene families, with expansions enriched for triterpene/sesquiterpene biosynthesis and (in A) cytochrome P450 3380/5358 (A); 2344/5971 (B)
  • Comparative analysis identified 227 inversions, 10867 translocations, and 5354 duplications between homologous chromosomes of subgenomes A and B 227 inversions; 10867 translocations; 5354 duplications
  • 69242 of 77266 genes (89.62%) contained 4788 Pfam domains; 151 domains (3.15%) were specific to subgenome A and 139 (2.90%) specific to subgenome B, a significant difference (P<0.05, Fisher's exact test) 151 vs 139 subgenome-specific domains
  • Significant difference in counts of retained P. stipuleanatus homologous genes between the two subgenomes P = 0.03417
Key statistics
  • count 3.45 Gb genome size, 24 chromosomes, 77266 protein-coding genes (final T2T genome assembly statistics)
  • other N50 contig length 147.37 Mb (assembly contiguity, vs 19.75 Mb in ginseng v1.0)
  • other BUSCO completeness 99.3% (genome), 98.4% (gene prediction) (assembly/annotation quality)
  • fold_change Subgenome A: 1.94 Gb / 40550 genes; Subgenome B: 1.51 Gb / 36716 genes (subgenome size and gene content comparison)
  • other Divergence time ~6.07 million years ago (subgenomes A/B); ~4.98 million years ago (subgenome B vs P. vietnamensis var. fuscidiscus) (phylogenomic divergence estimates)
  • other Ks peaks at 0.35-0.38 and 1.37-1.48 (within subgenomes, shared WGD events) and ~0.03 (between subgenomes A and B) (whole-genome duplication timing analysis)
  • pvalue P < 0.05 (Fisher's exact test for Pfam domain count differences between subgenomes)
  • pvalue P = 0.03417 (difference in retained P. stipuleanatus homologous gene counts between subgenomes A and B)

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 genome assembly and comparative genomics study of Panax ginseng, primarily descriptive/bioinformatic in nature (genome assembly metrics, phylogenomics, synteny, Ks-based divergence dating, gene family expansion/contraction). Formal inferential statistics appear only in a small number of comparisons of gene/domain counts between the two subgenomes, where Fisher's exact test was used to assess whether counts differed from expectation. Results for these comparisons are reported as p-values without accompanying dispersion measures, effect sizes, or stated multiplicity correction.

Replicationunclear Sample sizeNot described in terms of biological/technical replicates or power; comparisons are based on genome-wide counts of genes, Pfam domains, or homologous gene retention between the two assembled subgenomes GroupsSubgenome A vs subgenome B (gene/domain counts, homolog retention) Pairingna Randomization/blindingna Dispersionnone Exact p-valuesyes Effect sizesno Confidence intervalsno
Statistical tests used
Test Applied to n Assumptions
Fisher's exact test Comparison of gene counts associated with each Pfam domain between subgenome A and subgenome B (Fig. 3A) 69,242 of 77,266 total genes contained 4788 Pfam domains not stated
Fisher's exact test (implied; not explicitly named for this comparison) Comparison of counts of retained homologous genes of Panax stipuleanatus between subgenomes A and B (Fig. 3B) not stated
Approaches that could also have been used
  • Fisher's exact test was applied separately across a large number of Pfam domains (4788) to compare gene counts between subgenomes, with results summarized as significant/not significant.
    Could also: A multiple-testing correction such as Benjamini-Hochberg FDR — When the same type of test is repeated across thousands of features, an FDR or family-wise correction is a standard complementary approach that helps control the rate of false positives arising from testing many domains simultaneously.
  • Differences in gene/domain counts between subgenomes are reported via exact p-values (e.g., P<0.05, P=0.03417) without an accompanying effect-size measure.
    Could also: Reporting an odds ratio or relative risk with a 95% confidence interval alongside the Fisher's exact test — An effect-size estimate with a confidence interval would convey the magnitude and precision of the count difference between subgenomes, complementing the significance test.
  • Categorical count comparisons between subgenomes (e.g., retained homologous gene counts) were assessed with a single exact test.
    Could also: A chi-square test of independence, or a logistic/binomial regression framework — For larger contingency tables or when incorporating additional covariates (e.g., chromosome position, gene family), a chi-square test or regression-based approach can also be used and may allow modeling of multiple explanatory factors simultaneously.
  • Divergence times between subgenomes and related species were estimated from Ks peak locations and phylogenomic trees, presented as point estimates (e.g., ~6.07 million years ago).
    Could also: Reporting confidence or credible intervals around divergence time estimates (e.g., from Bayesian molecular dating methods) — Divergence time estimation methods that generate interval estimates (rather than single point estimates) can also convey the uncertainty inherent in molecular dating.

What was reproduced

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

Scope — pmid-38883331

Paper: Telomere-to-telomere reference genome for Panax ginseng highlights the evolution of saponin biosynthesis. Horticulture Research 11(6):uhae107 (2024). PMID 38883331 · PMCID PMC11179851 · DOI 10.1093/hr/uhae107.

Code artifact (P16 third-party tool): https://github.com/sc-zhang/T2Tools (t2tools.py with subcommands centel, gapcount, visual). This is the authors' cited tool for telomere / centromere identification and gap counting on the assembly. Per the brief, applying an existing third-party GitHub tool to the paper's own data is an equally valid reproduction.

Data (public, no auth):

  • Assembly FASTA (3.49 GB): figshare 10.6084/m9.figshare.25477741.v1 → Panax_ginseng_T2T_genome.fasta (https://ndownloader.figshare.com/files/46403986). HEAD-verified: 302 → signed S3, filename confirmed. This is the published T2T assembly — the exact input the paper's downstream tools ran on.
  • Raw reads (de novo assembly inputs): NCBI PRJNA752920 / PRJNA302556 (RNA-seq, ONT); CNCB PRJCA022032 (HiFi, Hi-C, Illumina). Not needed for the in-scope reproduction.
  • Annotation: figshare same record (gff3/pep/cds/gene fasta). Not in scope.

IN SCOPE (pipeline-derived, low-hanging — reproduce with T2Tools on the FASTA)

Run the paper's own tool (t2tools.py) on the published assembly FASTA and compare to the reported values. Compute is light (TRF over 24 chromosomes); the only large object is the 3.49 GB FASTA, which lives on «infra» only.

Claim Reported Paper loc Tool / method to reproduce
Telomeres identified 48 Table 1 / Results t2tools.py centel --telo_type plant → count telomere arrays (best_candidate_telo.list)
Centromeres identified 24 Table 1 / Results t2tools.py centel → best_candidate_centro.list
Gaps in final assembly T2T ⇒ 0 (286 gaps filled during assembly) Results t2tools.py gapcount → sum of N-runs across sequences
Assembly size 3.45 Gb Table 1 seqkit stats on FASTA (direct, sanity)
Number of chromosomes 24 Abstract/Table 1 count FASTA sequences (sanity)
Contig N50 147.37 Mb Table 1 seqkit stats (direct, sanity)
GC content 34.71 % Table 1 seqkit stats (direct, sanity)

Primary target = telomere count (48) and gap count (0) via the authors' own tool; assembly-stat rows are cheap direct cross-checks from the same FASTA.

Telomere motif: paper reports plant telomere (AAACCCT)n (= rev-comp TTTAGGG), which is exactly T2Tools centel --telo_type plant (searches TTTAGGG). Defaults align with the tool's documented behaviour, so no parameter guessing is needed for the telomere call.

OUT OF SCOPE (the hard ~20% — not attempted, by 80/20 rule)

  • De novo T2T assembly from HiFi/ONT/Hi-C (hifiasm + Hi-C scaffolding + TGS-GapCloser + Graph-Based Gap Filling on a 3.45 Gb genome) — heavy multi-day compute, not the cited-tool reproduction. Skipped: cost ≫ value; the assembly is published and is the input to the in-scope step.
  • Genome annotation (77,266 protein-coding genes; BUSCO 98.4 %; LAI 8.71) — separate annotation pipeline, out of scope.
  • BUSCO completeness (99.3 %) of the assembly — reproducible in principle (run BUSCO on the FASTA) but a separate tool, not T2Tools; treated as optional stretch, only if time permits after the telomere/gap reproduction.
  • Saponin / dammarenediol-synthase evolutionary analyses — wet-lab + bespoke comparative genomics, out of scope.

Reproduction strategy

All compute on «our HPC» (SLURM, partition std, no --mem). All data + repo on «infra» work dir «path». «host» keeps only small result values + pointers. One sbatch job: download FASTA → conda env (python, trf, matplotlib/numpy/pathos, seqkit) → git clone T2Toolscentel + gapcount → e

Figures / tables: Table
C1
Reported
48 telomeres
Reproduced
45 (T2Tools centel, all 24 chromosomes). The 3 not called are the 5' ends of Chr01/Chr05/Chr23, where the telomeric (AAACCCT)n motif IS present in the assembly but degraded/interrupted (longest perfect tandem run ~49 bp < the 70 bp / 10-copy stretch a single TRF record must contain), so T2Tools' strict TRF-based criterion misses them. A tool-independent terminal motif scan (telomere_terminal_scan.tsv) confirms ALL 48/48 chromosome ends carry telomeric arrays (>=10 heptamer copies in the terminal 20 kb) -> the paper's 48 is independently corroborated. The paper used quarTeT, which tolerates degraded arrays. Chr04 .dat obtained via windowed TRF (chr04_windowed_method.md): whole-Chr04 TRF exceeds the 12h cluster wall, so TRF was run on 2 Mb windows (identical params 1 1 2 80 5 200 2000) and offset-merged into one Sequence: Chr04 .dat -- exactly T2Tools' own --split large-genome path; telomere counts are provably window-safe (end-anchored).
within tolerance
C2
Reported
24 centromeres
Reproduced
24 (T2Tools centel best_candidate_centro.list = exactly 24 candidate regions, 1 per chromosome). Matches reported 24 exactly. max_pattern is computed genome-wide (Chr04 included); centromere stays >=1 region/chr under windowing so the count is window-safe.
exact
C3
Reported
0 gaps
Reproduced
exact
C4
Reported
3.45 Gb
Reproduced
3,455,427,034 bp (3.455 Gb)
exact
C5
Reported
24 chromosomes
Reproduced
24
exact
C6
Reported
147.37 Mb contig N50
Reproduced
147,374,239 bp (147.37 Mb)
exact
C7
Reported
34.71% GC
Reproduced
34.71%
exact
C8
Reported
99.3% BUSCO (assembly, embryophyta_odb10)
Reproduced
NOT COMPLETED (stretch goal, NOT part of the core in-scope T2Tools pipeline). Could not build a BUSCO env: shared /tmp is a 50MB tmpfs at 100% full and conda/python tempfile fail regardless of TMPDIR/TMP/TEMP redirection; /usw quota exceeded; «infra» conda inode-limited; apptainer unavailable on front1. Multiple distinct install routes attempted. Documented; separate tool, not the centel/gapcount pipeline in scope.
partial

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 92/100

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)
🤝
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.

1.7 M
tokens (I/O) · 123.2 M incl. cache
838 min
runtime · 127.32 CPU-h
34.8 GB
peak RAM
8 (7 failed)
HPC jobs
hummel
machine