A target enrichment method for gathering phylogenetic information from hundreds of loci: An example from the Compositae.
The main results reproduced: recomputed values matched the published ones within tolerance.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓Reported values were directly comparable
- ✓Reported values are derivable from the shared data
- ✓Any deviation was negligible
- ✓The central claim held under reproduction
- 🟡Could not use the authors’ exact input data
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡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
Reproduced the paper's downstream analytical pipeline (concatenated-alignment statistics and ML tree topology) from the paper's own Dryad-deposited derived data, recovered via an Internet Archive Wayback Machine workaround since the current Dryad platform blocks unauthenticated downloads. Taxon count (16) and parsimony-informative sites (28,324) matched exactly; alignment length matched within 0.06%; missing-data percentage was in the same range but off by ~5.4 points (59% reported vs 64.44% observed); the reconstructed ML tree topology matched the deposited GARLI tree exactly (RF=0/26) after harmonizing tip-label conventions, using IQ-TREE's ultrafast bootstrap+SH-aLRT as a documented, practically-necessary substitute for GARLI's computationally infeasible standard bootstrap. The upstream raw-read stage (trimming/assembly/ortholog assignment) could not be exercised because the paper's stated SRA BioProject PRJNA236448 does not contain retrievable reads matching this study (it is linked only to an unrelated 2019 SRA record) -- a genuine source-side data-availability gap, not an infrastructure limitation.
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.
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
- 2026-08-02
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-08-03no human curator yet
- Last updated
- 2026-08-03
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: opusCan a set of sequence capture probes targeting conserved orthologous sequences (COS) identified from Compositae ESTs, together with a bioinformatic/phylogenetic workflow, efficiently generate phylogenetically informative data from hundreds of nuclear loci to improve resolution of the Compositae base tree? The authors test whether such an approach recovers known relationships both across the family and among closely related species.
- ★ A custom sequence capture probe set (9678 baits targeting 1061 orthologous genes) was designed to enrich COS loci across the Compositae. resource
- ★ Application of the approach to 15 species yielded phylogenetically informative sequence data from 763 loci. finding
- ★ Maximum likelihood trees from the COS loci reconstruct known phylogenetic relationships across the Compositae family, mirroring the 10-locus chloroplast base tree. finding
- ★ A complete bioinformatic and phylogenetic workflow (PRINSEQ, BLAST prefiltering, Pairfq, Velvet/VelvetOptimiser, PHYLUCE, MAFFT, GARLI) was developed and made publicly available on GitHub. method
- ★ The majority of targeted loci are lost at the orthology assignment step in PHYLUCE rather than during hybridization/sequencing, due to its conservative rejection of putative paralogs. mechanism
- ★ Orthology detection is dependent on taxon sampling: restricting PHYLUCE to Heliantheae-only taxa yielded longer alignments for the same loci. finding
- The probe set also resolves relationships among closely related species (Helianthus and sister genus Phoebanthus) with high bootstrap support. finding
- Genomic DNA-based capture is preferable to transcriptome sequencing when fresh RNA is unavailable and enables use of museum specimens. method
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Targeted sequence capture (MYbaits target enrichment) followed by Illumina paired-end sequencing | 15 Compositae/Calyceraceae species (10 spanning the family plus 4 additional Helianthus and Phoebanthus tenuifolius) | hybridization enrichment with biotinylated 120-mer RNA baits (modified protocol: Hyb #1 reduced 20 µL to 15 µL; beads preincubated 30 min at room temperature with 5 µg denatured salmon sperm DNA) | captured COS locus sequence reads; number of COS loci recovered per species | MYbaits (MYcroarray); TruSeq DNA Sample Preparation kit (Illumina); Illumina HiSeq 2000, paired end, 100 base reads |
| Whole genome shotgun (WGS) sequencing of unenriched libraries | same 15 species | none (noncaptured/unenriched control libraries with unique TruSeq barcodes) | baseline percentage of reads with significant BLAST hits to COS loci, used as denominator for fold-enrichment | Illumina HiSeq 2000, paired end, 100 base reads |
| Genomic DNA extraction and library quantification | 15 species (greenhouse-grown, field-collected, and USDA accessions) | none | DNA yield/library concentration | DNeasy Plant Mini Kit (QIAGEN); Qubit 2.0 Fluorometer (Life Technologies) |
| BLAST-based ortholog identification / COS set construction | Helianthus annuus and Lactuca sativa ESTs vs. Arabidopsis single-copy genes; ca. 19,000 Carthamus tinctorius unigenes derived from ca. 41,000 ESTs | none (in silico) | number of COS loci with safflower alignments (E-value ≤ E-40, span ≥150 bp) | BLAST version 2.2.6 and 2.2.26; MUSCLE version 3.8 |
| Putative intron position determination for probe placement | Arabidopsis sequences from each COS alignment vs. full Arabidopsis genome database | none (in silico) | putative intron/splice-site positions mapped onto global multispecies alignments | BLAST; custom scripts from Chapman et al. (2007); arabidopsis.org database |
| Read quality trimming and de novo assembly | COS-DNA paired-end reads from each of the 15 species | none (bioinformatic) | number of contigs assembled per species; number of COS loci recovered as best BLAST hits | PRINSEQ version 0.18.2 (-min_len 40, -min_qual_mean 15, entropy lc_threshold 60, -trim_ns_right 10, -ns_max_p 20); Pairfq; Velvet (k = 51–99) with VelvetOptimiser version 2.2.0 |
| Orthology assignment and multiple sequence alignment | assembled contigs from 15 species plus lettuce whole genome assembly (version 4) sequences | none (bioinformatic); conservative rejection of contigs with more than one probe match as putative paralogs | number of orthologous COS loci aligned (763 across all taxa; 415 for Heliantheae-only); alignment lengths | PHYLUCE version 0.1.0 (match_contigs_to_probes.py, get_match_counts.py, get_fastas_from_match_counts.py, seqcap_align_2.py); LASTZ version 1.02.00; MAFFT version 7.029b |
| Maximum likelihood phylogenetic inference on concatenated data sets | 16 taxa (15 sampled species + lettuce); data sets of 763 loci, 186 loci (10/16 species), 347 loci (8/16 species), and Heliantheae-only 415 loci | none; varying locus/taxon completeness thresholds | tree topology and bootstrap support values | GARLI version 2.0, GTR-gamma model, 100 search replicates, 1000 bootstrap replicates |
- – 763 COS loci with sufficient coverage in a minimum of three species were aligned and used for phylogenetic analysis, out of 1061 targeted genes. 763 of 1061 loci
- – The 763-locus maximum likelihood tree recovered relationships mirroring the 10-chloroplast-locus base tree, with bootstrap support greater than 75% on almost all nodes. >75% bootstrap on almost all nodes
- – Fold-enrichment of COS-DNA over WGS-DNA varied across species, with Heliantheae taxa tending to show higher enrichment and the outgroup and most basal taxon showing lower values. 3–62 fold across species (Helianthus niveus subsp. tephrodes highest at 62; Phoebanthus lowest at 3)
- – No overwhelming trend was observed between fold enrichment and phylogenetic distance from the species used in bait design.
- ▲ Heliantheae-only PHYLUCE run produced 415 aligned loci with longer mean alignment length than the same loci in the all-taxa analysis. mean 404 bp (range 200–1101 bp) vs. mean 353 bp (range 27–1545 bp)
- – The 186-locus tree was congruent with the 763-locus tree; the 347-locus tree differed only in the placement of species within Heliantheae, suggesting a tradeoff between taxon representation and total data/number of genes.
- – The polyploid Senecio vulgaris did not recover fewer loci despite PHYLUCE's conservative paralog rejection. 316 loci
- ▼ Fulcaldea stuessyi, the most basal taxon, had the lowest number of BLAST-filtered reads, the lowest number of Velvet contigs, and the lowest number of COS BLAST hits—even lower than the outgroup. 174 loci
- count 763 COS loci; 269,585 bp; 59% missing data; 28,324 parsimony informative characters (Main concatenated maximum likelihood data set (15 sampled species + lettuce))
- count 9678 baits targeting 1061 orthologous genes (Final MYbaits probe set design; 120-mer baits tiled with 60-base overlap)
- count 186 loci; 49,918 bp; 37% missing data (Data set of loci represented in 10/16 species)
- count 347 loci; 96,649 bp; 45% missing data (Data set of loci represented in 8/16 species)
- count 415 loci; 167,650 bp (Heliantheae-only data set (Phoebanthus as outgroup))
- count 624 of ~1300 COS loci received safflower alignments (BLAST of ~19,000 Carthamus tinctorius unigenes against the ~1300 conserved gene set)
- fold_change fold-enrichment 3 to 62 (COS-DNA vs. WGS-DNA percentage of reads with significant BLAST hits to COS loci, per species (Table 1))
- other E-value ≤ E-40 and spanning ≥150 bp (Threshold for accepting best safflower BLAST hits into COS alignments)
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 is a methods/protocol paper describing development of a target-capture probe set for phylogenomic sequencing across the Compositae, rather than a hypothesis-testing experiment. Phylogenetic relationships were reconstructed using maximum-likelihood analysis (GARLI, GTR-gamma model) on several concatenated multi-locus alignments (763, 186, 347, and 415 loci), and confidence in the resulting tree topologies was assessed with nonparametric bootstrap resampling (1000 replicates) rather than classical significance testing.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Maximum likelihood phylogenetic reconstruction (GTR-gamma model of nucleotide substitution, 100 search replicates) | Concatenated alignments of 763, 186, 347, and 415 (Heliantheae-only) COS loci | 763 loci / 269,585 bp (59% missing data); 186 loci / 49,918 bp (37% missing data); 347 loci / 96,649 bp (45% missing data); 415 loci / 167,650 bp | not stated |
| Nonparametric bootstrap resampling | Nodal support on all reconstructed ML trees (Fig. 1B) | 1000 bootstrap replicates | not stated |
| BLAST similarity search with fixed E-value thresholds (e.g., E ≤ 1e-5, E ≤ 1e-40) | Read filtering, orthology/homology assignment, and fold-enrichment calculation | — | not stated |
-
Node support was assessed using nonparametric bootstrap resampling (1000 replicates) under a maximum-likelihood framework.↳ Could also: Bayesian phylogenetic inference (e.g., MrBayes, BEAST) reporting posterior probabilities — Posterior probabilities offer a complementary probabilistic measure of clade support and would let a reader compare confidence estimates from two different inferential frameworks.
-
Trees were built from concatenated multi-locus alignments (e.g., 763 loci) analyzed as a single supermatrix under one substitution model.↳ Could also: Coalescent-based species-tree methods (e.g., ASTRAL, *BEAST) that summarize individual gene trees — Summary-coalescent approaches explicitly accommodate gene-tree discordance from incomplete lineage sorting, which concatenation assumes away, and can be run alongside concatenation for comparison.
-
Orthology of captured loci was determined via BLAST best-hit criteria within PHYLUCE.↳ Could also: Reciprocal-best-BLAST-hit or explicit gene-tree-based orthology inference (e.g., OrthoFinder, phylogenetic orthology assignment) — Reciprocal or tree-based evidence adds an additional layer of confirmation for orthology calls beyond a single-direction best BLAST hit.
-
The effect of missing data was examined by comparing trees built from differently filtered locus sets (763-, 186-, 347-, 415-locus) and describing the differences narratively.↳ Could also: Formal topology tests (e.g., Shimodaira-Hasegawa test, approximately unbiased test) or quantitative tree-distance metrics (e.g., Robinson-Foulds distance) — These provide a quantitative measure of similarity or statistical support for topological differences between alternative trees, complementing visual/narrative comparison.
-
Rate variation and missing data were handled with a single GTR-gamma model applied uniformly across each concatenated alignment.↳ Could also: Partitioned analyses using model-selection tools (e.g., PartitionFinder) to assign locus- or codon-specific substitution models — Partitioning can capture heterogeneity in evolutionary rate or substitution pattern among loci, which a single unpartitioned model averages over.
-
Fold-enrichment for each species was reported as a single descriptive percentage of BLAST-matching reads.↳ Could also: Reporting enrichment with a dispersion measure (e.g., range or SD) if multiple technical replicates or sequencing lanes are available — A dispersion estimate alongside the point value would convey how consistent enrichment efficiency was, in addition to the single reported percentage.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
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 downstream analytical stage reproduced cleanly: taxon count (16) and parsimony-informative characters (28,324) matched exactly, alignment length within 0.06% (269,751 vs 269,585 bp), and an independent IQ-TREE GTR+G reconstruction recovered the authors' deposited GARLI topology at RF=0/26. Two deviations remain, both explainable and on our/definitional side: missing data 64.44% vs the reported ~59% (the paper never defines its denominator), and a substitution of UFBoot2+SH-aLRT for GARLI's computationally infeasible 1000 nonparametric bootstrap replicates, which makes support values — but not topology — incomparable. The genuine defect is on the authors'/repository side but in availability, not in the numbers: SRA PRJNA236448 contains none of this study's reads (only an unrelated 2019 record) and 2 of 4 Dryad files are unrecoverable, so the entire raw-read-to-alignment stage was untestable. Overall yellow: every reproducible claim held, but roughly half the declared pipeline could not be exercised at all.
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.