TC-hunter: identification of the insertion site of a transgenic gene within the host genome.
The main results reproduced, with only marginal, non-material deviations.
- Nothing in this column.
- 🟡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
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, healthy). Tool = TC_hunter Nextflow DSL1 @01096b2 (3rd-party tool on the paper's own data). (1) Tool sanity on shipped fly Test_data: EXACT - full 7-process pipeline ran, engineered insertion called at the injected position. (2) Host coverage of all 4 mouse WGS samples (PRJNA662713): real BWA-MEM alignment to GRCm38 on «our HPC» (BAMs 60-83 GB, 98.8-99.4% mapped) -> within-tol vs paper Table (base_count/genome +-1%; aligned meandepth -3..-5%, a conservative cross-check; per-sample ordering reproduces). (3) TIS host breakpoints corroborated independently by soft-clipped-read + mate-unmapped/discordant-pair pileups at each reported coordinate: M42 EXACT (32,944,479), M45 -1bp, M47 +20bp, M41 +54bp, all with 51-76 soft-clips - strong support for the reported insertion sites. (4) BLOCKED 1:1: per-sample construct coverage + exact TC_hunter TIS scores, because the 9.4 kb construct reference FASTA was never deposited (config points to a local-only path absent from all 5 author repos + SRA); only a feature table (T_scan_out: TH/Globin/Wip1/HSV/AmpR) survives in TC_hunter_Config-More. (5) OUT OF SCOPE: simulated-data sensitivity/precision/recovery (no deposited truth set) and wet-lab PCR validation. Honest verdict: described well enough to reproduce host coverage and the host-side insertion loci; the central construct-anchored numbers are blocked by a missing required input, not by method ambiguity.
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.
-
v1 current initial assessment Score 69assessed: 2026-06-22 ⛓ ad07050728cc
✎ 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
- 2026-06-22
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19no 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: sonnetChimeric reads and discordant read pairs from NGS data can be used to accurately and automatically identify the insertion site of a transgene within a host genome, addressing the lack of publicly available, well-documented tools for this purpose.
- ★ TC-hunter is an open-source Nextflow pipeline that identifies transgene insertion sites using chimeric reads and discordant read pairs from NGS data. resource
- ★ TC-hunter identifies transgene insertion sites with high sensitivity (98%) and precision (92.45%). finding
- ★ TC-hunter successfully identified and experimentally validated the transgene insertion site in four PPM1D-transgenic mouse samples. finding
- ★ A prediction score combining chimeric read count and discordant read pair count (weighted 1000x less) ranks candidate insertion sites by evidence strength. method
- TC-hunter generates circular plots and IGV snapshots to aid manual filtering and interpretation of candidate insertion sites. method
- TC-hunter requires only construct fasta, reference genome fasta, construct BED annotation, and sequencing data (fastq or bam) as inputs. method
- Genomic rearrangement (coverage decrease) at a predicted insertion site strengthens confidence in that prediction. mechanism
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Whole genome sequencing (WGS) | PPM1D-transgenic mouse DNA samples (M41, M42, M45, M47) | transgene overexpression (pronuclear injection of human PPM1D construct) | chimeric reads and discordant read pairs indicating transgene insertion site location | — |
| Read alignment (BWA MEM) | Mouse genome (GRCm38) + transgenic construct composite reference | none | mapped reads, CIGAR strings for chimeric/split alignments | BWA 0.7 / samtools 1.10 |
| Touchdown-PCR followed by Sanger sequencing | PPM1D-transgenic mouse DNA samples | none | experimental validation of predicted breakpoint sequence | — |
| Coverage analysis | PPM1D-transgenic mouse genome and construct | transgene insertion | average read coverage depth over host genome and construct regions | — |
- – TC-hunter achieved 98% sensitivity and 92.45% precision in identifying transgene insertion sites. 98% sensitivity, 92.45% precision
- – TC-hunter identified 8 total candidate TIS across 4 samples with scores between 1.000 and 16.051; the top-scoring candidate per sample was validated in all four samples. 8 TIS, scores 1.000-16.051
- – Sample M41's second candidate TIS (score 8.011) had high coverage and estimated ≥5 PPM1D copies but was not confirmed by PCR, suggesting a false positive from a repetitive region. score 8.011
- ▲ Construct coverage was similar to host coverage in M42 and M47, but 3.4x and 5.3x higher than host coverage in M41 and M45 respectively. 3.4-fold (M41), 5.3-fold (M45)
- ▼ Host coverage decreased by 66% near the validated TIS in sample M41, supporting a genomic rearrangement. 66%
- – Average sample processing after alignment used ~1.85 CPUs, 3.68G memory, in ~13.45 minutes. 1.85 CPUs, 3.68G, 13.45 min
- – Host genome average read coverage ranged from 33.48X to 59.53X across samples. 33.48X-59.53X
- other sensitivity 98%, precision 92.45% (overall TC-hunter performance metrics)
- count 8 predicted TIS across 4 samples (TIS candidates identified across all transgenic mice)
- fold_change construct coverage 3.4x higher than host (M41) (sample M41 construct vs host coverage)
- fold_change construct coverage 5.3x higher than host (M45) (sample M45 construct vs host coverage)
- other prediction score = number of chimeric reads + (number of discordant read pairs / 1000) (TIS ranking formula)
- mean host coverage 33.48X-59.53X (average read coverage over host genome across samples)
- count 26 discordant read pairs and 10 chimeric reads (supporting evidence for top M41 candidate TIS)
- other TC-hunter scores ranged 1.000-16.051 (range of prediction scores across all candidate TIS)
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 software methods paper presenting TC-hunter, a Nextflow-based bioinformatics pipeline for identifying transgenic insertion sites (TIS) from NGS data. No formal inferential statistical tests were performed; the primary quantitative evaluation consisted of sensitivity (98%) and precision (92.45%) computed against ground-truth validation by Touchdown-PCR and Sanger sequencing on four transgenic mouse samples. A heuristic prediction score combining chimeric read counts and discordant read pair counts was used to rank candidate sites; all other results (coverage, candidate counts) were reported descriptively.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Sensitivity and precision (binary classification performance metrics) | Validation of TC-hunter TIS predictions against Touchdown-PCR / Sanger sequencing results across 4 transgenic mouse samples (8 total predicted TIS; 4 top-ranked candidates experimentally tested) | 4 transgenic mice; 8 predicted TIS total | na |
| Heuristic scoring formula (chimeric reads + discordant read pairs / 1000) | Ranking of all predicted TIS candidates within each sample | — | na |
-
Sensitivity (98%) and precision (92.45%) were reported as single point estimates derived from n=4 samples and 8 predicted TIS candidates↳ Could also: Exact (Clopper-Pearson) or Wilson score confidence intervals for sensitivity and precision could also be reported — With a small validation set, point estimates carry substantial uncertainty; CIs would make that uncertainty explicit and allow readers to gauge the reliability of the quoted performance figures
-
False-positive TIS candidates were distinguished from true positives by manual inspection of circular plots and IGV snapshots, guided by the prediction score↳ Could also: A receiver operating characteristic (ROC) or precision-recall curve varying the score threshold could also characterize tool performance objectively across the full score range — A threshold-sweep analysis would reveal the sensitivity-specificity trade-off at every possible cutoff, helping future users select a score threshold appropriate for their own data and tolerance for false positives
-
Tool performance was assessed exclusively on four samples carrying a single transgenic construct (PPM1D) in mouse↳ Could also: Benchmarking on a broader set of samples (different constructs, host species, sequencing depths, or coverage levels) or on simulated data with known ground truth could also be performed — A more diverse benchmark would allow assessment of how sensitivity and precision generalize across varying construct architectures, genome assemblies, and sequencing parameters
-
TC-hunter was compared to alternative tools descriptively via a supplementary feature table (Table S1) listing availability and documentation↳ Could also: A quantitative head-to-head benchmark on the same four samples against available comparable tools could also be conducted — Running alternative tools on identical data would allow direct numerical comparison of sensitivity and precision rather than only feature availability, giving readers a clearer basis for tool selection
-
The prediction score weights chimeric reads 1000× over discordant read pairs using a fixed, manually chosen ratio↳ Could also: A logistic regression or penalized linear model trained on labeled TIS data could also be used to derive data-driven weights — An empirically fitted model could optimize the relative weighting and, with sufficient training data, provide calibrated probabilities rather than an arbitrary-scale composite score; the heuristic approach is transparent and requires no training data, which is a practical advantage in low-n settings
-
Per-sample read coverage over the host genome was summarized as a simple average (e.g., 33.48X–59.53X across samples)↳ Could also: Reporting coverage distribution per sample (e.g., median ± IQR or mean ± SD across genomic windows) could also convey uniformity of coverage — A single mean coverage value does not distinguish uniformly covered genomes from those with high variance across regions; distributional summaries would help readers assess whether uneven coverage might affect TIS detection in a given sample
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-35184734 (TC-hunter, BMC Genomics 2022)
Paper: Börjesson V et al. "TC-hunter: identification of the insertion site of a transgenic gene within the host genome." BMC Genomics 23:149 (2022). DOI 10.1186/s12864-022-08376-0 · PMCID PMC8859905.
Tool / code: TC_hunter — Nextflow pipeline.
- Brief lists
github.com/bcfgothenburg/SSFwhich points to the real repogithub.com/vborjesson/TC_hunter(commit pinned: 01096b259fa784080eeecbc7aa13d22a70dc3cce, 2023-03-14). - Two entry points:
TC_hunter_BWA.nf(FASTQ in → BWA MEM align → call TIS) andTC_hunter.nf(BAM in → call TIS). Deps: R≥3.5, python2.7, samtools 1.10, nextflow 19.01.0, bwa 0.7. Conda env shipped:Scripts/TC_hunter.yml. - Method: chimeric reads (one read maps to both construct & host) + discordant read pairs (mate maps to construct, other to host). MAPQ threshold default 30. Score = (#chimeric reads) + (#discordant pairs / 1000).
Data: SRA BioProject PRJNA662713. 32 runs total; only 4 WGS runs are the TC-hunter samples (the other 28 are RNA-Seq from a separate expression study — OUT of scope here). WGS runs (Mus musculus, Illumina WGS):
- SRR17772246 = PPM1D_M41 · SRR17772245 = PPM1D_M42
- SRR17772244 = PPM1D_M45 · SRR17772243 = PPM1D_M47 Transgene: human PPM1D construct inserted into mouse genome.
IN SCOPE (pipeline-derived, attempted)
- Pipeline sanity run — run TC_hunter on the repo's shipped
Test_data/(JointRefGenome.fasta + construct.txt). Confirms the tool runs end-to-end and producesoutput_summary.html. (Not a paper number, a reproducibility gate.) - Per-sample raw coverage (host) — derivable two ways:
(a) cheap: deposited base_count ÷ mouse genome size (already consistent, see
AUDIT); (b) exact: from BWA-aligned BAM via
samtools depth/mosdepth. Paper Table: M41 49.53, M42 44.34, M45 49.40, M47 33.48 (host). - Construct coverage per sample (M41 166.26, M42 35.67, M45 260.81, M47 21.68) — from BAM over the construct contig. Needs construct ref.
- Predicted Transgene Insertion Sites (TIS) per sample — the central claim: M41 chr16:62,428,722–62,428,726 (score 10.026) + chr9 candidate; M42 chr16:32,944,479–32,974,010 (15.035); M45 chr9:74,912,357–74,969,077 (16.051) + 3 minor candidates; M47 chr5:23,254,639–23,254,658 (9.039). Reproduced by running TC_hunter_BWA on each WGS run vs mouse ref + PPM1D construct, comparing top-ranked breakpoints + scores.
CONDITIONALLY IN SCOPE (needs an input we must locate)
- TIS calls require the construct reference FASTA (human PPM1D transgene
construct sequence) +
construct.txtfeature table + construct length. If the exact construct is NOT deposited/derivable, items 3–4 cannot be reproduced 1:1 → would be recorded as a blocker, not fabricated.
OUT OF SCOPE (not pipeline-derived from deposited data, NOT attempted)
- Simulated-data benchmark: sensitivity 98.00%, precision 92.45%, 81.67% TIS recovery — built from in-house simulated Drosophila+Orc6 insertions; the simulated reads/insertion truth set are not deposited → cannot reproduce 1:1.
- External validation on soy (ST77-KP2) and rice (T1c-19, TT51) samples — these belong to other studies' accessions; secondary, attempt only if time permits.
- Wet-lab validation (PCR/Sanger confirmation of insertion sites) — wet-lab, out.
- 28 RNA-Seq runs in PRJNA662713 — belong to a different analysis, not TC-hunter.
Compute plan
All heavy steps on «our HPC» (SLURM, «infra»). Downloads on front1 into
«path».
Per-sample WGS fastq ~62–87 GB compressed (4 samples ≈ 312 GB). BWA MEM align to
mouse ref + construct, then TC_hunter TIS calling. Pull back only small outputs
(output_summary tables, coverage numbers, TIS coordinates).
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.
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-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.