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

Free circular introns with an unusual branchpoint in neuronal projections.

Elife · 2019
L1 99/100 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.

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
Every question reproduced
-1 pts
From: “every question reproduced”
Total score -7
✓ What held up
  • Same input data as the authors
  • Reported values were directly comparable
  • No relevant deviation in data/preprocessing
  • 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)
  • Every checked point held up.
How its reproducibility compares
99/100
Reproducibility score
1.4 SD above mean
vs. all fields · 1173 studies
🎯 Scores higher than 95% of all assessed papers rank 55 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

salvaged by watchdog from agreement.json (agent omitted ROOM_RESULT.json)

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
2026-06-19
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-18
no human curator yet
Last updated
2026-07-29

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: opus
Founding hypothesis

Which RNAs—particularly intron sequences (both retained and excised)—are localized to neuronal projections, as systematically identified by sequencing total (rRNA-depleted) RNA rather than polyadenylated RNA from primary rat hippocampal neurons.

Core claims
  • A set of free circular introns with a non-canonical (C) branchpoint is enriched in distal neuronal projections; these appear to be tailless lariats that escape debranching. finding
  • These free circular introns lack ribosome occupancy, sequence conservation, and known localization signals, and their function is unknown. finding
  • Total RNAseq of physically dissected neuro-glial projections versus whole cells reliably distinguishes and quantifies projection-localized versus nuclear-localized RNAs. method
  • Hundreds of intron regions (retained and excised) are localized to projections, and 1632 intron regions show reliable coverage in projections. finding
  • Contrary to prior reports, no Kcnma1 introns are retained or localized to projections; only spliced Kcnma1 isoforms are detected in projections. finding
  • Ribosomal protein mRNAs and mitochondrial transcripts are enriched in neuronal projections. finding
  • A culture system using semipermeable membranes with 1 μm pores physically separates neuro-glial projections from cell bodies/nuclei for compartment-specific RNA sequencing. method
  • Sequenced datasets (total RNAseq, PASseq, ribosome profiling, polyA+ RNAseq) constitute a resource for studying RNA localization in neurons (GSE129924). resource
Experimental setups
Assay System Perturbation Readout Platform
rRNA-depleted total RNAseq (paired-end) primary rat hippocampal neuro-glial cultures on 1 μm pore membranes (projection and whole cell) DNA replication inhibitor to block glial division; physical fractionation RNA abundance (TPM) of polyA+ and polyA- long RNAs; intron region coverage Illumina, 100–125 nt reads, ~200 nt insert
PolyA-site sequencing (PASseq) primary rat hippocampal cultures (projection and whole cell) none polyadenylation sites (to distinguish retained introns from intronic polyA)
Ribosome profiling cytoplasmic (nuclei-depleted) fraction of primary rat hippocampal neurons on plates none ribosome occupancy / coding exon identification
PolyA+ selected RNAseq cytoplasmic (nuclei-depleted) fraction of primary rat hippocampal neurons on plates none coding exon identification
single-molecule FISH (smFISH, RNAScope) primary rat hippocampal neurons none subcellular localization of mRNAs, exons, and introns RNAScope probe sets
Immunofluorescence imaging primary rat hippocampal neuro-glial cultures on membranes none MAP2/DAPI/GFAP/Vimentin localization confirming projection separation
Microcapillary electrophoresis (Bioanalyzer RNA pico) total RNA from whole cells and projections none RNA size distribution Bioanalyzer
Key results
  • 1440 transcripts significantly enriched in projections, including Pabpc1, Map2, Dlg4, Gfap fold-change >1.5, q<0.01
  • 1486 genes significantly depleted from projections, including nuclear ncRNAs Xist, Malat1, Meg3, snoRNAs, scaRNAs >1.5 fold, q<0.01
  • 70 annotated ribosomal protein mRNA isoforms enriched in projections >2-fold
  • Of 190,180 intron regions, 57,432 reliably covered in whole cells but only 1632 in projections 1632 of 190,180
  • Classification of 1632 projection intron regions: 385 unannotated alt 5'SS/polyA (high EI), 320 alt 3'SS/TSS (high IE), 428 retained introns (high EI+IE)
  • smFISH confirmed Creld1 free intron detectable in distal projections, not colocalizing with Creld1 exons
  • No Kcnma1 introns retained/localized to projections; only spliced isoforms detected
  • Calm2 intron-retaining isoform abundant in both compartments with no selective enrichment
Key statistics
  • count 190,180 intron regions total (genomic intron regions analyzed)
  • count 57,432 reliably covered in whole cell; 1632 in projection (intron region detectability across 5 biological replicates)
  • count 1440 enriched, 1486 depleted in projections; 16,899 no significant enrichment (differential expression of 19,815 transcripts)
  • correlation Spearman ≥0.83 (projection), ≥0.88 (whole cell) (replicate TPM correlation)
  • fold_change >1.5 fold, q<0.01 (significance threshold for projection enrichment/depletion)
  • fold_change >2-fold (70 RP mRNA isoforms enriched in projections)
  • count 30–80 million mate pairs per sample (total RNAseq depth, 5 replicates/10 samples)
  • other projection:whole cell TPM — Pabpc1 40:12, Rpl4 297:96, Srsf5 18:37, Ubc 162:146, Ppib 43:50, Polr2a 2:4 (smFISH-validated mRNA ratios)

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 study physically dissected neuronal projections from whole cells in primary rat hippocampal cultures and compared RNA content using rRNA-depleted total RNA-seq (n=5 biological replicates per condition). Transcript abundance was quantified with Kallisto and differential expression between projections and whole cells was tested with Sleuth, using a combined q-value <0.01 and fold-change >1.5 significance threshold. Intron-region coverage was assessed via fixed count-density thresholds rather than a formal statistical test, and selected genes were validated by smFISH. Replicate reproducibility was summarized using Spearman's correlation on TPM values.

Replicationbiological Sample sizeFive biological replicates for total RNAseq (projection and whole cell); three biological replicates each for PASseq; three biological replicates for ribosome profiling and polyA+ RNAseq from plate cultures. No formal power analysis described. GroupsNeuronal projections (underside of semipermeable membrane) versus whole cells (top surface), from mixed neuro-glial primary rat hippocampal cultures Pairingunpaired Randomization/blindingnot stated Dispersionnone Exact p-valuesno Effect sizesno Confidence intervalsno Multiplicity correctionFDR q-value (Sleuth default; Benjamini-Hochberg implied)
Statistical tests used
Test Applied to n Assumptions
Sleuth differential expression test (Wald test with FDR q-value) on Kallisto TPM quantifications Projection versus whole-cell comparison across 19,815 annotated RNA transcripts 5 biological replicates per condition (10 samples total) not stated
Spearman's rank correlation coefficient Pairwise biological replicate reproducibility within whole-cell and projection RNAseq datasets (TPM values) 5 biological replicates per condition not stated
Gene ontology enrichment analysis (specific test not named) Genes significantly enriched in projections or whole cells not stated
Fixed coverage threshold filtering (≥1 read per replicate and mean read density >0.005 reads/nt) 190,180 intron regions screened for reliable coverage in projection and whole-cell libraries 5 biological replicates na
Approaches that could also have been used
  • Differential expression was assessed using Kallisto pseudo-alignments fed into Sleuth
    Could also: DESeq2 or edgeR (via tximport to import Kallisto count estimates) could also be used for projection vs. whole-cell differential expression — DESeq2 and edgeR are extensively benchmarked negative-binomial frameworks for small-n RNA-seq; running them alongside Sleuth would provide a sensitivity check on findings, as the three methods can yield different results for low-count features
  • Intron-region enrichment in projections was identified using fixed read-density thresholds rather than a formal statistical test
    Could also: A count-based differential analysis on intron bins (e.g., DEXSeq, or featureCounts + DESeq2) could also formally test enrichment across the 190,180 intron regions with FDR correction — A model-based approach would provide calibrated error rates across the full set of tested intron regions, complementing the threshold-filtering strategy
  • Replicate reproducibility was summarized pairwise using Spearman's correlation on TPM values
    Could also: Principal component analysis (PCA) or multidimensional scaling on sample-level counts could also be used for quality control — PCA simultaneously visualizes variance structure and potential outliers across all samples, providing a global view that pairwise correlations summarize but do not display
  • Results are reported as fold-change values without accompanying uncertainty estimates
    Could also: Shrinkage-based log-fold-change estimation (e.g., DESeq2 lfcShrink with apeglm) could also be reported alongside nominal fold-changes — Shrinkage estimators reduce variance inflation for low-count features, producing more stable rankings of enriched transcripts and making effect sizes more comparable across the abundance range
  • The gene ontology enrichment test method and background gene set were not specified
    Could also: Standard approaches such as a hypergeometric test or Fisher's exact test with FDR correction (e.g., clusterProfiler, topGO, or g:Profiler) with an explicit background could also be named — Specifying the background set and correction method allows readers to assess the scope of the multiple-testing correction applied across GO terms and to reproduce the enrichment analysis
  • smFISH validation was assessed qualitatively by visual inspection of spot distributions in cell bodies versus projections
    Could also: Quantitative per-cell spot counting in soma versus projection compartments, summarized with a ratio and tested with a paired or mixed-effects model, could also be used — Quantitative spot counts would provide a numerical enrichment estimate with dispersion, enabling a direct statistical comparison between smFISH-derived ratios and RNAseq TPM projection:whole-cell ratios
Software: Kallisto · Sleuth · TopHat2 · RepeatMasker

Result convergence & founder nodes

Findings this paper shares with others that ran a comparable experiment. A node’s strength is how many independent papers report it (replication breadth) — not how often it is cited, so a heavily-replicated but under-cited founder still stands out.

Citation network

Where this publication sits in the reproducibility-weighted citation graph — what it is built on, and what is built on it. Citation data from OpenAlex.

Citations
18
Impact: medium
Foundation confidence
None of its references are in our reproducibility record yet — its foundation cannot be assessed.
Topics

No assessed neighbours yet — the network grows as more papers are assessed.

Data lineage

The datasets this paper uses (text-mined from the full text via Europe PMC), and which other assessed papers stand on the same data. A shared dataset is a factual link — not a judgement.

GSE129924 GEO in Article (http://semanticscience.org/resource/SIO_001029)
no other assessed paper uses this yet
NR_132635.1 RefSeq in Methods (http://purl.org/orb/Methods)
no other assessed paper uses this yet

What was reproduced

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

Scope — pmid-31697236

Paper: Saini H, Bicknell AA, Eddy SR, Moore MJ. "Free circular introns with an unusual branchpoint in neuronal projections." eLife 2019;8:e47809. PMCID PMC6879206 · DOI 10.7554/eLife.47809.

Listed code: https://github.com/marvin-jens/find_circ (a third-party tool — find_circ, Memczak 2013 — used by the authors, per P16 equally valid). Data: GEO GSE129924 / SRA PRJNA533159 (22 samples; Rattus norvegicus). Author supplement (key): http://eddylab.org/publications/Saini19/Saini19-supplement.tar.gz — ships the analysis code (R notebooks, Perl/Python) AND intermediate inputs (kallisto abundances for all 10 total-RNAseq samples; source-data tables for Figs 2 & 3). This makes most of the pipeline-derived results reproducible deterministically without re-aligning raw reads.

Pipelines used in the paper

  1. kallisto 0.44.0 (transcript quant, Rnor_6.0.91 cDNA+ncRNA+Xist, 100 boots, --rf-stranded) → sleuth 0.30.0 (R 3.5.1) DE: whole-cell (CB) vs projections (P), Wald test on conditionP. → Figure 2.
  2. Custom intron-region pipeline (TopHat2/Bowtie2 align to Rnor_6.0 Ensembl 81; exon2intron_dexseqgff.pl + dexseq_prepare_annotation.py to define 190,180 intron regions; per-region coverage, exon-intron (EI) / intron-exon (IE) junction reads, PASseq polyA reads, ribosome-profiling reads) → intron classification (Figs 3–4). Counting logic is in Rscript_plots_Saini_et_al_2019.Rmd operating on the shipped Figure3-SourceData1_All_Introns.tsv (190,180 rows × 73 cols).
  3. find_circ (Memczak 2013; bowtie2 → unmapped2anchors → find_circ.py) on the projection total-RNAseq → circularly-permuted (circular intron) junction reads; per-intron circ_num_p column feeds Fig 5. Branchpoint analysis (Fig 6) from RT mismatches in circularly-permuted reads.
  4. cleanUpdTseq (PASseq polyA-site calling) — supporting.

In scope (pipeline-derived → attempted)

# Result Pipeline How reproduced
A1 Fig 2 DE table + projection-enriched / -depleted transcript counts kallisto→sleuth re-run sleuth 0.30.0 on shipped kallisto outputs; cross-check vs shipped Fig2 source data
A2 190,180 intron regions; 1,632 P-covered; 57,432 CB-covered; EI/IE classes 385/320/428/499; free 499→96 polyA / 221 ribo / 278 free custom intron pipeline (downstream classification) recompute deterministically from shipped All_Introns source table using the exact Rmd filter logic (validates numbers are derivable from shipped data → anti-fabrication check)
B circular introns in projections (circ_num_p≥2, enriched) + branchpoint count (14; 12 C / 2 G) find_circ on projection RNA-seq run find_circ on the 5 projection total-RNAseq libraries (SRR8919670–674); compare detected circular-intron set to the All_Introns circ_num_p column and the paper's branchpoint result

Out of scope (not pipeline-derived / not attempted)

  • Wet-lab: RNase-R/RNase-H validation, RT-PCR, Northern blots, branchpoint sequencing of individual introns, microscopy (Figs 1, 6A/B, 7).
  • Manual curation steps (e.g. manual selection of RP genes from GO:0005840; manual branchpoint inspection of the 14 introns).
  • GO-term enrichment (Fig 3/4 supplements) — depends on an external GO service snapshot; low priority.
  • PASseq cleanUpdTseq polyA-site calling — supporting, raw-data heavy; not primary.

Reproduction strategy

A2 first (fast, deterministic, confirms ~10 reported numbers). A1 next (genuine re-run of the statistical pipeline from shipped intermediates). B last (heaviest; the named repo tool on raw reads). 80% floor = A1+A2; B is the harder reach.

Figures / tables: Fig3Fig4Fig 2AFig 6
C1
Reported
190180
Reproduced
190180
exact
C2
Reported
57432
Reproduced
57432
exact
C3
Reported
1632
Reproduced
1632
exact
C4
Reported
1599
Reproduced
1599
exact
C5
Reported
33
Reproduced
33
exact
C6
Reported
385
Reproduced
385
exact
C7
Reported
320
Reproduced
320
exact
C8
Reported
428
Reproduced
428
exact
C9
Reported
499
Reproduced
499
exact
C10
Reported
96
Reproduced
96
exact
C11
Reported
221
Reproduced
221
exact
C12
Reported
278
Reproduced
278
exact
C13
Reported
1440
Reproduced
1440
exact
C14
Reported
1486
Reproduced
1488
within tolerance
C15
Reported
14
Reproduced
14
exact
C16
Reported
12C/2G
Reproduced
m.public.grade.uncheckable

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 99/100

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.

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
Every question reproduced
-1 pts
From: “every question reproduced”
Total score -7

This is essentially a 1:1 computational reproduction: the authors shipped both their analysis code and intermediate inputs, and 14/16 claims reproduce exactly — counts recomputed deterministically from the All_Introns table and the DE counts (1440) re-run in sleuth with cor(b)=cor(qval)=1.0. The only numeric deviation is C14 (1488 vs 1486, 0.13%), two transcripts flipping across the q=0.01 cutoff under R 3.6.3 vs 3.5.1 — a technical version effect, not a methodology or authors' defect. The one flagged anomaly (C11's mislabeled # 221 code comment) is a stale comment, not fabrication, since the reported 96/221/278 partition is internally consistent and fully derivable; C16 (12C/2G branchpoint) is simply out of computational scope (wet-lab RT-mismatch). Overall quality is green.

🤝
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.

346.6 k
tokens (I/O) · 39.2 M incl. cache
64 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.