Base editing in human cells with monomeric DddA-TALE fusion deaminases.
The main results reproduced: recomputed values matched the published ones within tolerance.
- 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 (1:1 on the core ND1/ND4/ND6 results). Pipeline=MAUND amplicon base-editing quantifier (ibs-cge/maund @6b5d90c) plus an independent from-first-principles per-position C->T profiler as cross-check, both run on raw SRA fastq (PRJNA805019) on «our HPC» SLURM («job», all 57 sample-rows = ND1 9 constructs + ND4/ND6 5 constructs each x3 reps). The three abstract headline maxima reproduced essentially exactly: ND1 42.0% (reported ~42%), ND4 31.0% (~31%), ND6 27.3% (~27%). ND1 L-GSVG per-position profile (8 target C positions, Fig3f) reproduced within <=0.5 points at every position. Untreated background ~0%, construct activity ranking reproduced. ND4/ND6 peaks match; some secondary per-position values differ due to reconstructed-amplicon window/strand. Paper is described well enough to reproduce from raw data. NOT attempted (same method, deprioritized): MT-TC Fig4, AAV time-course Fig5, whole-mtDNA off-target Fig6, nuclear Fig1-2.
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 ⛓ 2d3c4c95d9a5
✎ 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
- 2026-06-21
- 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: sonnetThe authors test whether non-toxic, full-length variants of the interbacterial toxin DddA_tox can be engineered and fused to a single TALE array (rather than split between two TALE arrays) to create monomeric DdCBEs (mDdCBEs) that overcome the delivery and targeting limitations of conventional dimeric split-DddA_tox base editors.
- ★ mDdCBEs built from non-toxic full-length DddA_tox variants enable mitochondrial DNA base editing with efficiencies of up to 50% upon transient expression in human cells finding
- ★ Non-toxic, full-length DddA_tox variants (AAAAA, E1347A, GSVG) were engineered via structure-based site-directed mutagenesis and random error-prone PCR mutagenesis method
- ★ mDdCBEs delivered via AAV achieve nearly homoplasmic (>99%) C-to-T mtDNA editing in cultured human cells finding
- ★ mDdCBEs often produce mutation patterns at target sites that differ from those produced by conventional dimeric DdCBEs finding
- ★ mDdCBEs allow base editing at genomic sites where only a single TALE-binding sequence can be designed, unlike dimeric DdCBEs finding
- ★ Transfection of mDdCBE-encoding mRNA rather than plasmid DNA reduces off-target editing in human mitochondrial DNA finding
- Within the GSVG variant, the G1348S mutation is essential for non-toxicity, S1326G is neutral, and A1398V/S1418G reduce cytotoxicity mechanism
- ★ Base edits induced by non-toxic full-length DddA_tox fusions are stably maintained in cells for at least 21 days, indicating lack of cytotoxicity finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| targeted deep sequencing | HEK293T cells | transfection with structure-based DddA_tox variant (AAAAA, E1347A) fused to dCas9/D10A nCas9 + UGI | C-to-T base editing frequency and indel frequency at TYRO3 site | — |
| targeted deep sequencing | HEK293T cells | transfection with random-mutagenesis-derived GSVG DddA_tox variant fused to N/C termini of Cas9, D10A nCas9, H840A nCas9, dCas9 | C-to-T base editing and indel frequency at TYRO3 and other nuclear sites | — |
| clonal isolation and genotyping | HEK293T cells | transfection with AAAAA or E1347A variant fusions | proportion of clonal populations carrying targeted C-to-T edits at TYRO3 and ROR1 site 1 | — |
| targeted deep sequencing | HEK293T cells (mtDNA) | mDdCBEs containing GSVG or E1347A variant fused to TALE arrays targeting ND4, ND6, ND1 | mitochondrial C-to-T editing frequency vs. conventional split DdCBE pairs | — |
| targeted deep sequencing | mouse NIH3T3 cells (mtDNA) | mDdCBE transfection targeting MT-ND5 gene | C-to-T editing frequency | — |
| targeted deep sequencing | HEK293T cells (mtDNA) | mDdCBE (GSVG) targeting MT-TC site with only one TALE-binding sequence | C-to-T editing frequency at a site inaccessible to dimeric DdCBEs | — |
| AAV2 transduction followed by targeted deep sequencing | HEK293T cells (mtDNA) | AAV2-delivered mDdCBEs targeting ND4 and ND1 at variable viral doses/time points | time- and dose-dependent C-to-T editing frequency | AAV2 vector |
| mitochondrial genome-wide high-throughput sequencing | HEK293T cells (mtDNA) | split DdCBE pairs vs. mDdCBEs (E1347A/GSVG) vs. TALE-free DddA_tox constructs vs. no treatment | genome-wide off-target C-to-T editing frequency | — |
- ▲ AAAAA variant fusion induces C-to-T conversions immediately upstream of the protospacer at TYRO3 with frequencies up to 43% 43%
- ▲ E1347A variant fusion retains residual deaminase activity, editing at 37% (nCas9 fusion) or 16% (dCas9 fusion) at the same position 37%/16%
- ▲ GSVG variant fusions induce C-to-T conversions at various nuclear sites with efficiencies of up to 38% 38%
- ▲ mDdCBEs containing GSVG achieve mtDNA editing at ND4, ND6, and ND1 up to 31%, 27%, and 42% respectively, on par with dimeric DdCBE pairs up to 42%
- ▲ AAV2-delivered mDdCBEs reach editing frequencies of 99.1% at ND4 and 59.8% at ND1 with high multiplicity of infection 99.1%
- ▲ mDdCBE targeting the single-TALE-binding-site MT-TC gene achieves 16% C-to-T editing, a site not targetable by dimeric DdCBEs 16%
- – Base edits induced by non-toxic DddA_tox fusions are maintained in cells for at least 21 days without loss
- – TALE-free DddA_tox constructs show mitochondrial genome-wide off-target editing (0.018-0.019%) similar to negative control (0.019%), whereas split DdCBEs show elevated off-target editing (0.031-0.19%)
- fold_change 43% (AAAAA variant editing frequency at TYRO3 C-3 position)
- fold_change up to 50% (mDdCBE mtDNA editing efficiency upon transient expression)
- fold_change 99.1% (AAV-delivered mDdCBE editing at ND4 site at high MOI)
- count 9/11 clones (81%) (clonal editing at TYRO3 site with AAAAA fusion)
- count 8/17 clones (47%) (clonal editing at ROR1 site 1 with AAAAA fusion)
- other 0.019% (average genome-wide mtDNA off-target C-to-T editing frequency, negative control)
- fold_change ~2-fold decrease (GSAG/GSVS revertant editing frequency decline from day 3 to day 21 post-transfection)
- other 8.4% (proportion of TC motifs in human mtDNA editable by mDdCBE but not DdCBE)
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.
The paper reports base-editing and indel frequencies at targeted genomic and mitochondrial sites as measured by targeted deep sequencing, summarizing results as means ± s.e.m. from three independent experiments per condition (bar graphs and heat maps across Figs. 1-6 and supplementary figures). Comparisons between constructs (e.g., monomeric vs. dimeric DdCBEs, different DddA variants, AAV dose/time-course, clonal editing persistence) are presented descriptively, based on the reported frequencies and heat-map patterns, without any formal hypothesis test, p-value, or multiplicity correction described in the text.
-
Editing frequencies across constructs (e.g., mDdCBEs vs. dimeric DdCBE pairs, different DddA variants) are compared descriptively via bar graphs and heat maps, without a formal statistical test or p-value.↳ Could also: A two-way ANOVA (construct × site) or a mixed-effects model with post-hoc comparisons — This would provide a formal estimate of whether differences in editing frequency between constructs exceed what would be expected from experiment-to-experiment variability, complementing the visual comparison of bar graphs.
-
Variability across three independent experiments is summarized using the standard error of the mean (s.e.m.).↳ Could also: Reporting the standard deviation (SD) or a 95% confidence interval alongside or instead of s.e.m. — At small n (n=3), s.e.m. can visually understate the spread among replicates; SD or a CI often conveys the actual variability and precision of the estimate more directly for readers.
-
The three independent experiments underlying each mean are not explicitly labeled as biological or technical replicates.↳ Could also: Explicitly designating replicate type (e.g., independent transfections/independent cell passages) in the methods or figure legends — Clarifying replicate type would help readers judge the level of independence captured by the error bars and the appropriate interpretation of the reported variability.
-
Many site/construct/timepoint comparisons are made in parallel across Figs. 1-6 and the supplementary figures without a stated correction for multiple comparisons.↳ Could also: A false discovery rate procedure (e.g., Benjamini-Hochberg) or a Bonferroni-type correction applied across the family of comparisons — Such a correction could help control the overall false-positive rate when evaluating many editing-frequency comparisons together, if formal significance testing were introduced.
-
Clonal editing outcomes are described as raw proportions (e.g., 9 of 11 clones edited for one variant vs. 16 of 21 for another) without a formal statistical comparison between variants.↳ Could also: A Fisher's exact test or chi-square test comparing the proportion of edited clones between the AAAAA and E1347A fusion variants — This would yield a p-value or effect size for whether the clonal editing rates differ between variants, supplementing the direct comparison of counts.
-
The persistence of editing over time (day 3 to day 21) is described narratively (e.g., an approximate two-fold decrease) rather than with a formal trend test.↳ Could also: A repeated-measures ANOVA or linear mixed model across timepoints within the same samples — This would formally quantify whether editing frequency changes significantly over time, accounting for repeated measurements from the same replicate samples.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-35821233 (Mok et al. 2022, Nat Commun; monomeric DddA-TALE base editors)
Pipeline-derived results (IN SCOPE)
The only computational/bioinformatic pipeline in the paper is MAUND (github.com/ibs-cge/maund, Code Availability), which quantifies base-editing and indel frequencies from targeted amplicon deep sequencing (SRA PRJNA805019, Illumina MiniSeq). Reported editing efficiencies (% reads with C->T conversion) and indel frequencies are all MAUND outputs. In scope:
- Fig.3 (a-f): mtDNA C->T editing efficiency + per-position editing for ND1, ND4, ND6 by monomeric (L-/R-GSVG, -E1347A, -E1347A_G1348S) and split (G1333/G1397) DdCBEs vs untreated. (Headline: GSVG up to ND1 42%, ND4 31%, ND6 27%.)
- Fig.4 (c,d): MT-TC (tRNA-Cys) editing (~16% GSVG).
- Fig.5: AAV-delivered editing time-course (ND1/ND4).
- Fig.6 / Suppl.: whole-mtDNA-wide off-target C->T (Whole-MTseq).
- Fig.1,2: nuclear (TYRO3 etc.) DddA-nCas9 editing/indels.
Primary reproduction target
Fig.3 mtDNA editing efficiencies for ND1 (all 9 constructs) + headline GSVG/E1347A for ND4, ND6, and MT-TC (Fig.4) — the core claims. Reproduce by running MAUND (authors' tool) and an independent per-position profiler on the raw SRA fastq, comparing edited/total and per-position C->T to the paper's Source Data (MOESM6) and abstract.
OUT OF SCOPE (not attempted / wet-lab / external)
- Protein/plasmid design, in vitro deaminase assays (Suppl Fig 1), cloning, transfection, cell culture, AAV production, Sanger — wet-lab.
- mitoTALEN/off-target structural interpretation — not pipeline-derived.
- Whole-mtDNA-wide off-target (Fig.6) and AAV time-course (Fig.5): same MAUND pipeline, deprioritized for time; reproducible by the same method (noted, not fully run).
Method/parameters
Nested sequencing amplicons reconstructed from rCRS (NC_012920.1) + Suppl. Table 4 target-specific primers. rgen (target spacer) per gene/side derived from the edited-C cluster; target_nt=C; window positions 4-11 (Methods). Reads joined (R1/R2 overlap) before MAUND. Compute on «our HPC» SLURM; data on «infra».
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.