MirDIP 5.2: tissue context annotation and novel microRNA curation.
The main results reproduced: recomputed values matched the published ones within tolerance.
- ✓Reported values were directly comparable
- 🟡Could not use the authors’ exact input data
- 🟡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
mirDIP 5.2 is an NAR Database resource paper; headline results are database-production totals. Reproduced by auditing the publicly deposited database (registration-free ZIPs at ophid.utoronto.ca/mirDIPweb) against the paper's reported counts, downloading + decompressing + counting on «our HPC» («infra»). RESULT: 7 of 9 headline numeric claims match the deposit EXACTLY -- 46,364,047 predictions, 27,936 genes, 2,734 microRNAs, 32,497 novel miRNAs, 28,557 unique novel sequences, 2,657 tissue-context miRNAs, 27,576 tissue-context genes. The 2 DERIVED/CURATED totals are not exactly recomputable from shipped artifacts: C6 (123,651,910 gene-miRNA-tissue interactions) -- best-effort co-expression recompute gives 1.53B and no score-class cutoff or the bidirectional record count (131,517,089) reproduces it; the exact aggregation rule is in the authors' unshipped tissue post-processing. C7 (330 contexts) -- a curated 209 normal+92 disease+29 other classification vs 326 miRNA-tissue contexts in the deposit. No fabrication indicator: all directly-countable numbers match exactly; C6/C7 reflect a transparency gap (aggregation code not shipped). De-novo regeneration of the predictions/tissue pipeline was NOT attempted (repo lacks workflow files, configs, conda env, and all raw inputs; hardcoded cluster paths -> docs_insufficient for regeneration). Scaffold's code link (NCI-Thesaurus/thesaurus-obo-edition) was a wrong auto-enrichment; real repo = github.com/ijlab/mirdip. The 5 GEO input expression series are all open, Homo sapiens, correct modality, full series present.
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 50assessed: 2026-06-19 ⛓ 437cd285388f
✎ 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
- 2026-06-29
- 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: sonnet- ★ mirDIP 5.2 removed eight outdated resources, added miRNATIP, and ran five prediction algorithms against miRBase and mirGeneDB miRNAs to expand and improve interaction coverage method
- ★ mirDIP 5.2 is the first database to provide miRNA-gene interaction predictions using mirGeneDB data resource
- ★ 32,497 novel microRNAs (28,557 unique sequences) were curated and integrated from 14 publications to accelerate use of novel miRNA data resource
- ★ mirDIP 5.2 associates tissue and disease context with microRNAs, genes, and microRNA-gene interactions using data from 20 resources resource
- The database enables searching by precursor ID and integrates miRAnno, a network-based pathway-association tool, plus a new API for programmatic access method
- ★ In disease conditions miRNAs appear more tissue-specific (expressed in fewer contexts), whereas in normal tissues more miRNAs are expressed across multiple tissues finding
- Cross-referencing novel miRNA sequences from Ali et al. against mirDIP identified more overlapping novel miRNAs across studies than the original publication found finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| in silico miRNA target prediction (miranda, BiTargeting, PITA, RNAhybrid, MirMAP) | human miRNAs (miRBase v.22, mirGeneDB 2.0) vs Ensembl 3' UTRs (GRCh38 release 103) | none | predicted miRNA-gene interactions and integrated interaction score | miranda, BiTargeting, PITA, RNAhybrid, MirMAP |
| small RNA sequencing (miRNA-seq) processing | human tissue/disease samples from GEO-derived paired datasets | none | miRNA expression (binary presence/absence and quantile-normalized abundance) | nf-core/smrnaseq v2.0.0 |
| bulk RNA sequencing (mRNA-seq) processing | human tissue/disease samples from GEO-derived paired datasets | none | gene expression (binary presence/absence) | nf-core/rnaseq v3.8.1 |
| sequence overlap/comparison analysis | novel miRNA sequences from 14 curated publications vs known miRBase/mirGeneDB miRNAs | none | number of overlapping miRNA sequences between novel and known/curated sets | — |
| literature curation via PubMed search | human novel microRNA publications | none | number of novel miRNAs and genomic coordinates collected | — |
| context/ontology mapping | tissue, cell type, and disease terms across integrated datasets | none | standardized context terms mapped via Disease Ontology, BRENDA Tissue Ontology, and OLS ontologies | Ontology Lookup Service (OLS) |
- – mirDIP 5.2 includes 46,364,047 predictions for 27,936 genes and 2734 microRNAs 46,364,047 predictions
- – 32,497 novel miRNAs (28,557 unique sequences) curated from 14 publications 32,497 novel miRNAs
- – 680 novel miRNA sequences already present in miRBase or mirGeneDB 680 sequences
- – Context annotation covers 330 tissue/disease contexts, 2657 miRNAs, 27,576 genes, and 123,651,910 gene-miRNA-tissue interactions 123,651,910 interactions
- – miRNA tissue expression collected for 301 contexts (209 normal, 92 disease) covering 2656 miRNA IDs 301 contexts
- – Gene expression context collected for 278 contexts (188 normal, 90 disease) covering 27,576 genes 278 contexts
- – In disease conditions miRNAs are expressed in as few as 1 up to 92 contexts, indicating greater tissue specificity than in normal tissue 1-92 contexts
- ▲ For 13 novel miRNAs from Ali et al., mirDIP identified 7 overlapping miRNAs with other studies (vs. 4 found by the original authors), 3 of which appear in 5-7 other papers 7 vs 4 overlaps
- count 46,364,047 (total miRNA-gene predictions in mirDIP 5.2)
- count 2734 (microRNAs included in mirDIP 5.2 predictions)
- count 27,936 (genes included in mirDIP 5.2 predictions)
- count 32,497 (novel microRNAs curated from 14 publications)
- count 123,651,910 (gene-microRNA-tissue interactions with context information)
- count 330 (tissue and disease contexts collected across miRNAs and genes)
- percentage 53% (of ontology relationships obtained via Disease Ontology and BRENDA Tissue Ontology)
- count 680 (novel miRNA sequences overlapping with known miRBase/mirGeneDB miRNAs)
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 paper describes the construction and update of mirDIP, a database that aggregates microRNA-gene interaction predictions from multiple existing prediction algorithms and resources, and integrates tissue/disease expression context from 20 externally collected and re-processed datasets. The work is primarily a data-integration and curation effort rather than a hypothesis-testing study: interactions are combined into an 'integrated score' based on rank/percentile across resources, and expression data are converted to binary (expressed/not-expressed) or five-level quantile-normalized categories per context. No inferential statistical tests (e.g., group comparisons, p-values) are reported; results are presented as counts, overlaps, and descriptive distributions (e.g., number of miRNAs per tissue).
-
MiRNA and gene expression values were converted to binary (expressed/not-expressed) calls per context, based on any non-zero read count in any replicate.↳ Could also: A statistical expression-calling approach (e.g., a read-count or TPM threshold combined with a variance-stabilizing or negative-binomial model, as used in DESeq2/edgeR) could also be used to define 'expressed' status. — Threshold- or model-based calling can help distinguish genuine low-level expression from technical noise and can propagate a measure of confidence into the binary call, which a simple non-zero rule does not capture.
-
Non-zero miRNA expression values were quantile-normalized within each sample and then discretized into five percentile-based classes (mirDIP-Tissues scale).↳ Could also: Retaining continuous quantile-normalized values, or using a rank-based non-parametric summary (e.g., median with interquartile range) across replicates, could also be used to represent expression level. — Discretization into five bins simplifies interpretation and querying, while continuous or rank-based summaries would preserve more of the original quantitative variation for downstream analyses that require finer resolution.
-
Evidence for a miRNA's expression in a context is scored as the number of sources reporting expression divided by the number of sources measuring that miRNA/context (a simple proportion).↳ Could also: A weighted meta-analytic combination (e.g., inverse-variance weighting by sample size, or Fisher's/Stouffer's method for combining independent evidence) could also be used to combine evidence across datasets. — Weighting by dataset size or precision can account for the fact that datasets vary substantially in sample size and platform, giving more influence to larger or more reliable studies rather than treating each source equally.
-
The interaction 'integrated score' is derived from combining multiple prediction algorithms and assigning score classes based on top-percent targets per miRNA across resources.↳ Could also: A formal ensemble/meta-prediction statistical framework (e.g., logistic regression, rank aggregation methods such as RobustRankAggreg, or a Bayesian evidence-integration model) could also be used to combine predictions from multiple algorithms. — Model-based aggregation can provide a probabilistic interpretation of the combined score and allow explicit weighting of algorithms by their individual predictive performance against a validation set.
-
Ontology term mapping (Disease Ontology, BRENDA Tissue Ontology, and additional OLS ontologies) was performed with a portion (27%) resolved through manual curation.↳ Could also: A quantitative inter-curator agreement statistic (e.g., Cohen's kappa) between the mapping curators and verifier could also be reported for the manually curated portion. — Reporting an agreement statistic is a standard way to convey the reliability/reproducibility of manual curation steps in a database, complementing the description of the curation workflow.
-
Novel miRNA sequence overlaps between publications (e.g., Ali et al. vs. other curated papers) are reported as raw overlap counts (Figure 3).↳ Could also: An enrichment statistic (e.g., a hypergeometric or Fisher's exact test comparing observed overlap to that expected by chance given the total curated miRNA universe) could also be used to characterize these overlaps. — A formal enrichment test would quantify whether the observed overlap between independently curated miRNA sets is greater than expected by chance, complementing the raw overlap counts already presented.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-36453996 (mirDIP 5.2)
Paper: Hauschild AC, Pastrello C, Ekaputeri GKA, et al. MirDIP 5.2: tissue context annotation and novel microRNA curation. Nucleic Acids Res 2023;51(D1):D217–D225. DOI 10.1093/nar/gkac1070 · PMID 36453996 · PMCID PMC9825511.
What kind of paper: an NAR Database issue resource paper. The "results" are database-production statistics (totals of integrated predictions, miRNAs, genes, novel miRNAs, tissue/disease contexts and interactions) plus a deposited, publicly downloadable database. It is not a hypothesis-testing analysis with a single deposited raw-data → figure pipeline.
Artifacts located (corrected from scaffold metadata)
- Code (authors'): https://github.com/ijlab/mirdip (branch
for_paper, pushed 2025-02-05, public, no license file). The scaffold'sgithub.com/NCI-Thesaurus/thesaurus-obo-editionlink is a wrong auto-enrichment and is ignored. - Deposited database (the paper's own data): direct, registration-free ZIPs at
https://ophid.utoronto.ca/mirDIPweb/:mirDIP_Unidirectional_search_v_5_2.zip(681 MB) — integrated predictionsmirDIP_Bidirectional_search_v_5_2.zip(1.8 GB)mirDIP_Novel_Annotation_v_5_2.zip(684 KB) — novel miRNA annotationmirDIP_Tissue_v_5_2.zip(51 MB) — tissue-context interactions (TSV)mirDIP_Tissue_v_5_2_novel.fasta.zip(215 KB) — novel miRNA FASTA
- Input expression datasets (tissue annotation): GEO GSE134949 (Rahman/SEAweb), GSE149084, GSE181922, GSE137308, GSE126448 + TCGA (gdac.broadinstitute.org) + GTEx (gtexportal.org). 20 resources/publications in total.
- Prediction inputs: miRBase v22, mirGeneDB 2.0; 5 algorithms run for them (miRanda, BiTargeting, PITA, RNAhybrid, MirMAP), plus ~16 integrated source resources (TargetScan, miRDB v6, RNA22, miRzag, miRNATIP, …).
What the repo actually ships (reproducibility surface)
prediction_update_and_integration/scripts/: per-toolsave_*_properly.pynormalizers, miRBase/gene-ID mappers (R/Py), benchmark R scripts, a Noisy-OR integration R script (mirdip5_run_noisyOR.R).mirdip-tissue/sources/: per-sourcedataCleanup*.Rmdnotebooks (+ rendered.nb.htmlwith outputs) and merge scripts;data/mirbase/mature_homo-sapiens_dataframe.txt.- MISSING for end-to-end replication: the Nextflow workflow files (
map_ids.nf, the nf-core run wrappers),nextflow.config/ijclusterprofile, themirbaseconverter.ymlconda env (referenced, not present), and all raw/ intermediate input data (prediction-tool raw outputs, the gold-standardplat_large_three_cols_only.tsv, benchmark collections, the per-source count matrices). Hard-coded absolute paths («path»,«path») confirm it was run on the authors' cluster and not packaged for portability.
IN SCOPE (pipeline-derived, attempted)
The honest, clear 1:1 target is auditing the deposited database against the
paper's headline production totals — i.e. does the public deposit actually
contain what the abstract/results claim? This is both a reproduction of the
central numeric claims and the delivers_promised profiling check, on the paper's
own data, using the paper's own deposit.
| # | Reported result | Pipeline / source | Reproduction method |
|---|---|---|---|
| C1 | 46,364,047 predictions | integration (Noisy-OR over ~16 resources) | count records in Unidirectional deposit |
| C2 | 27,936 genes | integration | count distinct genes in deposit |
| C3 | 2,734 microRNAs | integration | count distinct miRNAs in deposit |
| C4 | 32,497 novel miRNAs | novel curation (14 pubs) | count entries in Novel Annotation deposit |
| C5 | 28,557 unique novel sequences | novel curation | count unique sequences in novel FASTA |
| C6 | 123,651,910 gene–miRNA–tissue interactions | tissue annotation (nf-core rnaseq/smrnaseq + post-proc) | count records in Tissue deposit |
| C7 | 330 tissue/disease contex |
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.