Engineered ADARs Enable Precision A-to-G Base Editing of DNA

• p. 2 Results, Operating concept of snuABE

  strated superior editing activity at both the FANCF target (0.7%) and the AAVS1 safe-harbor locus (6.0%) (Fig. 1d)

• p. 2 Fig. 1d scatter plot, labelled data point

  P. colchicus

• p. 2 Results, 'Operating concept of snuABE using ADAR and a dedicated gRNA'

  demonstrated superior editing activity at both the FANCF target (0.7%) and the AAVS1 safe-harbor locus (6.0%)

Only the authors can say what is right here.

Judge's note: Read the sentence on crop C004-a and the Fig. 1d scatter plot on crops C004-b and C003-a plus the full page main-p002.png. The y axis is FANCFA12–G conversion (%) with ticks 0, 0.5, 1.0, 1.5; the x axis is AAVS1 A12–G conversion (%) with ticks 0–8. The P. humanus point sits at roughly x=6.0, y=0.73, matching the quoted 6.0% and 0.7%. The point labelled P. colchicus sits at roughly x=0.8, y=1.02, clearly above P. humanus on the FANCF axis. P. humanus is plainly the best ortholog at AAVS1 but not at FANCF, so "superior editing activity at both" is not supported by the panel it cites. I found no statement elsewhere defining a combined or ranked criterion that would make the claim hold.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 3 Results, 3′-end protection of tagRNA

  The highest editing efficiency was observed when La(1–194) was fused to the C terminus, reaching 16.5% for AAVS1 and 8.5% for FANCF

• p. 3 Fig. 2b (AAVS1), x-axis tick labels

  -La(Int)

• p. 3 Fig. 2c (FANCF), x-axis tick labels

  -La(C)

• p. 3 Fig. 2b,c bars

  -La(N) -La(Int) -La(C)

Only the authors can say what is right here.

Judge's note: Read the text on crops C007-a and C007-e and examined Fig. 2 on crops C007-b, C007-c, C007-d and the full page main-p003.png. Both panels have nine bars matching nine ticks in the order snuABE1.0, snuABE2.0, snuABE2.1, +Lafull, +La1–194, −La(N), −La(Int), −La(C), snuABE3.1 (the last three sit over the bracket labelled snuABE2.0, in the same N/Int/C order as the schematics in Fig. 2a). Reading against the axis gridlines, Fig. 2b gives approximately 0, 6.0, 7.6, 8.9, 9.0, 6.4, 17.2, 2.1, 20.2; Fig. 2c gives approximately 0.6, 3.8, 5.2, 6.2, 5.0, 3.3, 8.5, 2.8, 12.0. The last bar matches the text's snuABE3.1 values of 20.0% and 12.1% exactly, confirming the bar-to-tick alignment, and the text's 16.5%/8.5% match the seventh bar, labelled −La(Int), not the eighth bar labelled −La(C). Fig. 2a independently designates snuABE2.0-La(C) as snuABE3.0. As published the figure labels and the text disagree about which fusion position is best; nothing elsewhere in the article resolves which is correct.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 4 Results, Characterization of optimized conditions of snuABE3.1

  The highest editing efficiencies were consistently observed when adenines were located between positions 12 and 18

• p. 4 Results, same paragraph, two sentences later

  further supporting a confined editing window spanning positions 12–16

• p. 9 Discussion

  whereas snuABE enables editing within the editing activity window (positions 11–17)

• p. 4 Results, 'Characterization of optimized conditions of snuABE3.1'

  editing efficiencies were consistently observed when adenines were located between positions 12 and 18

Only the authors can say what is right here.

Judge's note: Read all three statements on the page images: crops C010-a/C010-d/C010-e/C010-i/C010-j and C010-b/C010-f on p.4 (full page main-p004.png), and crops C010-c/C010-g/C010-h on p.9. The wording is verbatim as quoted. Against the data, Fig. 3d on main-p005.png plots A8–A20 with the curves peaking at A14–A16 and falling to near-background at A11 and A19–A20, and Extended Data Fig. 2 (main-p016.png) gives maximum on-target values of 1.74% at A11, 4.51% at A12, 17.61% at A14, 17.48% at A15, 8.30% at A16, 3.88% at A17 and ~0.1–0.3% at A18. The Results' two ranges differ from each other and the Discussion's 11–17 corresponds to neither; no definition or threshold is given anywhere that would reconcile them. A reader takes away a different window depending on which passage is read, so only the authors can say which range is intended.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 4 Results, Characterization of optimized conditions of snuABE3.1

  cells were treated with snuABE3.1 using 32 tagRNAs targeting adenines at different positions

• p. 4 Fig. 3 caption, panel d

  d, Effective targeting range of snuABE across 13 distinct sites within the AAVS1 locus.

• p. 16 Extended Data Fig. 2 caption

  a-m, Editing efficiency of snuABE3.1 at target positions A8 to A20 across thirteen independent sites.

• p. 4 Results, editing window

  using 32 tagRNAs targeting adenines at different positions

Only the authors can say what is right here.

Judge's note: Read the Results sentence on crops C011-a and C011-d (full page main-p004.png), the Fig. 3d legend on crops C011-b, C011-e, C011-f, and the Extended Data Fig. 2 caption on crops C011-c and C011-g. I then counted the panels on the Extended Data Fig. 2 page image (main-p016.png): panels a–m are AAVS1 8A, 9A, 10A, 11A, 12A, 13A, 14A, 15A, 16A, 17A, 18A, 19A, 20A — thirteen sites — and every panel has rows 0 nt (sgRNA), 24 nt, 27 nt, 30 nt and 40 nt (43 nt in panel l), i.e. four tagRNAs per site. Fig. 3d on main-p005.png likewise plots A8–A20 with four tagRNA-length curves. Neither 13 nor 52 is 32. The number 32 belongs to the separate sequence-context experiment described in the next paragraph ("Using 32 target sites with varied 5′ and 3′ flanking sequences") and in the Fig. 3e legend ("at 32 endogenous loci"). Which count the authors meant for the window experiment cannot be settled from the files.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 23 Reporting summary, Life sciences study design, Replication

  All experiments using cell lines were independently replicated at least twice, and key findings were confirmed in three or more independent experiments, except for the in vivo N.C group, which was performed once (n = 1).

• p. 4 Fig. 3 caption, panel j

  j, Whole-RNA sequencing analysis of A-to-I editing ratios for ABE8e, H840A, snuABE1.0 and snuABE3.1, each performed in a single replicate.

• p. 6 Fig. 4 caption, panel k

  k, Whole-RNA sequencing analysis of A-to-I editing ratios for ABE8e, H840A, snuABE3.1 and snuABE4.1, each performed in a single replicate.

Only the authors can say what is right here.

Judge's note: Read page image p023 (the Replication box, quoted verbatim above), crop C039-b (Fig. 3 caption: 'j, Whole-RNA sequencing analysis of A-to-I editing ratios for ABE8e, H840A, snuABE1.0 and snuABE3.1, each performed in a single replicate... All experiments were performed in HEK293T cells.') and page image p006 (Fig. 4 caption: 'k, Whole-RNA sequencing analysis of A-to-I editing ratios for ABE8e, H840A, snuABE3.1 and snuABE4.1, each performed in a single replicate'; 'Unless otherwise indicated, experiments were performed in HEK293T cells'). These are cell-line experiments with n = 1, so the Reporting Summary's single stated exception (the in vivo NC group, also flagged n = 1 in the Fig. 5 legend) does not cover them. Only the authors can say how the Replication statement should be amended.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 13 Methods, In vivo delivery of snuABE through hydrodynamic tail-vein injection

  Mice were maintained under specific-pathogen-free conditions and used at 6–8 weeks of age.

• p. 13 Methods, In vivo delivery of snuABE through hydrodynamic tail-vein injection (next paragraph)

  were injected into 8–10-week-old C57BL/6 mice through the tail vein

• p. 24 Reporting summary, Laboratory animals

  Mice were maintained under specific pathogen-free conditions and used at 6–8 weeks of age.

• p. 13 Methods, In vivo delivery

  used at 6–8 weeks of age

Only the authors can say what is right here.

Judge's note: Opened crops C031-a and C031-b and the full page image p013: both sentences are in the subsection 'In vivo delivery of snuABE through hydrodynamic tail-vein injection', in consecutive paragraphs, and the only in vivo experiment described is the hydrodynamic injection. Page image p024 (Reporting summary, 'Laboratory animals') states 'C57BL/6J mice (Mus musculus) were purchased from Orient Bio (Republic of Korea). Mice were maintained under specific pathogen-free conditions and used at 6–8 weeks of age.' No acclimatisation interval or second cohort is stated anywhere in the subsection, so the two age ranges cannot both describe the injected animals; only the authors can say which is correct.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 2 Fig. 1d, data-point label

  M. drosophila

• p. 12 Methods, Cloning of snuABE constructs

  The Drosophila ADAR sequence was obtained from the HyperADAR control plasmid

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: Drosophila is a genus name, so it cannot serve as a species epithet in a binomial of the form "M. drosophila". Verified on the page image (crop C003-a and main-p002.png) that Fig. 1d prints "M. drosophila" alongside the properly formed abbreviations "P. colchicus", "P. humanus", "H. sapiens" and "D. opalescens". The Methods (crop C003-b, p.12) refer only to "the Drosophila ADAR sequence", and no organism with the epithet "drosophila" is named anywhere in the article. Only the authors can supply the intended binomial.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 8 Discussion, second paragraph

  protecting the 3′-end of tagRNA with La protein (as in PE7) contributed to the creation of the snuABE3 variants27,38

• p. 10 Reference list, entries 27 and 28

  28. Yan, J. et al. Improving prime editing with an endogenous small RNA-binding protein. Nature 628, 639–647 (2024).

• p. 3 Results, '3′-end protection of tagRNA increases snuABE editing efficiency'

  protecting the 3′-end of tagRNA might further enhance editing efficiency28

• p. 10 Reference list

  Yan, J. et al. Improving prime editing with an endogenous small RNA-binding protein.

Judge's note: Read crop C024-a (Discussion, p. 8) and the reference list crops C024-b/-d plus the layout text: ref. 27 = "Chen, P. J. et al. Enhanced prime editing systems by manipulating cellular determinants of editing outcomes. Cell 184, 5635-5652 (2021)", ref. 28 = "Yan, J. et al. Improving prime editing with an endogenous small RNA-binding protein. Nature 628, 639-647 (2024)", ref. 38 = "Anzalone, A. V. et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature 576, 149-157 (2019)". Internally, the paper uses ref. 27 for PEmax ("The additional mutations and NLSs used in PEmax were incorporated into snuABE2.1 (ref. 27)", same paragraph), ref. 38 for the additional-nick/PE3 idea (p. 6, "enhance editing efficiency (Fig. 4e)38"), and ref. 28 for the La/3'-end protection idea (crop C024-c, p. 3: "protecting the 3'-end of tagRNA might further enhance editing efficiency28"). The La/PE7 half of the Discussion sentence is therefore attached to refs 27,38 instead of ref. 28.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 4 Fig. 3 caption, panel i

  i, Comparison of A-to-I editing effects on seven endogenous mRNA sites using full-length hADAR1, hADAR2, snuABE1.0 and snuABE3.1.

• p. 4 Results, off-target effects on transcriptome

  first analyzed six well-characterized endogenous ADAR hotspot sites in human mRNAs

• p. 5 Fig. 3i heat map column headers

  PODXL

• p. 4 Fig. 3 legend, item i

  Comparison of A-to-I editing effects on seven endogenous mRNA sites

Judge's note: Read the Fig. 3 legend on crops C009-a, C009-d, C009-h and C009-j, the Results sentence on crops C009-b, C009-e, C009-g and C009-i, and counted the Fig. 3i heat map columns on crop C009-c and the full page main-p005.png. The heat map has exactly six labelled columns — AZIN1, CCNI1, COPA, GLI1, IGFBP7, PODXL — with five editor rows (nCas9-H840A, Human ADAR1, Human ADAR2, snuABE1.0, snuABE3.1), and the grid contains six cells per row. The legend's "seven" matches neither the panel nor the body text. I found no supplementary site counted in addition; the legend describes panel i specifically. Data are unaffected, so this is a caption count error.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Fig. 4 caption, panel g

  Each data point in g individually represents the mean of each 32 targets presented in Extended Data Fig. 2a,b.

• p. 16 Extended Data Fig. 2 caption

  Editing efficiency of snuABE3.1 at target positions A8 to A20 across thirteen independent sites

• p. 21 Extended Data Fig. 7 caption

  a,b, Editing rates of target adenines across all 16 possible NAN motifs at A12 a and A15 b, using snuABE3.1, snuABE3.1n, snuABE4.1, and snuABE4.1n.

• p. 6 Fig. 4 legend, item g

  individually represents the mean of each 32 targets presented in Extended Data Fig. 2a,b

Judge's note: Read the Fig. 4 legend on crops C015-a, C015-d, C015-f, C015-h and the full page main-p006.png. Checked the two candidate targets on the page images: Extended Data Fig. 2 (main-p016.png, crops C015-b/C015-k/C015-l) panels a and b are the AAVS1 8A and 9A tagRNA-length heat maps for snuABE3.1 alone, containing no 32-target data and none of the nicked or evolved variants; Extended Data Fig. 7 (main-p021.png, crops C015-c/C015-m) panels a and b plot all 16 NAN motifs at A12 and at A15 for snuABE3.1, snuABE3.1n, snuABE4.1 and snuABE4.1n — exactly the 32 targets and the four editors shown in the Fig. 4g violin plot (main-p007.png). The Results text for the same comparison cites "(Fig. 4g and Extended Data Fig. 7a,b)" (crops C015-e, C015-g). The legend pointer is wrong; the data are fine.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Fig. 4 legend g

  presented in Extended Data Fig. 2a,b

• p. 6 Results text

  (Fig. 4g and Extended Data Fig. 7a,b)

• p. 6

  Extended Data Fig. 2a,b. h, Fold changes in editing

Judge's note: Read crops C020-a, C020-c (Fig. 4 legend g, p. 6) and C020-b (Results, p. 6: "snuABE4.1n achieving a median editing efficiency of 5.4% (Fig. 4g and Extended Data Fig. 7a,b)"). Checked the Extended Data captions in the layout text: Extended Data Fig. 2 is "Efficiency comparison of snuABE3.1 at different target adenine positions. a-m, Editing efficiency of snuABE3.1 at target positions A8 to A20 across thirteen independent sites" (snuABE3.1 only, panels a-m), whereas Extended Data Fig. 7 is "Motif-wide editing activity and the effect of ngRNA positioning. a,b, Editing rates of target adenines across all 16 possible NAN motifs at A12 a and A15 b, using snuABE3.1, snuABE3.1n, snuABE4.1, and snuABE4.1n" - i.e. 16 x 2 = 32 targets for exactly the four editors in the violin plot. The legend pointer is wrong; it should read Extended Data Fig. 7a,b.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 20 Extended Data Fig. 6, structure labels (top-right panel and lower-right inset)

  E367F

• p. 20 Extended Data Fig. 6 caption

  The AI-predicted mutations I536S and E637F are indicated in magenta.

• p. 6 Results, AI-based directed evolution

  boring three substitutions (E438Q, I536S, E637F) showing the highest

• p. 6 Results, AI-based evolution

  E438Q, I536S, E637F

Judge's note: Read the full page image p020: the two right-hand panels both carry the label 'E367F' (the left-hand panels are labelled 'I536S'), and the caption reads 'The AI-predicted mutations I536S and E637F are indicated in magenta.' The substitution is E637F everywhere else I checked: Results p006 ('one variant harboring three substitutions (E438Q, I536S, E637F)') and the Fig. 4e construct schematics on p007 ('PhADARdd E438Q/Δ10aa, I536S; E637F'). No residue 367 appears anywhere in the article, so the digits are transposed in the figure labels.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Fig. 5b, bar-graph legend keys

  snuABE4.2/4.2n

• p. 9 Fig. 5b, heat map row labels

  snuABE4.1n

• p. 8 Results, 'Precise adenine correction in genetic disease models by snuABE'

  snuABE4.1n achieved bystander-free, pinpoint editing with an efficiency of up to 63.6% (Fig. 5b)

Judge's note: Read crop C028-a (key: blue "snuABE3.1/3.1n", red "snuABE4.2/4.2n", hatched "Indel"), crop C028-b (heat map rows snuABE3.1, snuABE3.1n, snuABE4.1, snuABE4.1n, and the bar graph with +/- rows for snuABE, tagRNA and ngRNA) and crop C028-c (Results p. 8). The bar graph has one bar per heat-map condition and the tallest red bar reads about 64%, matching the 63.6% the text attributes to snuABE4.1n; the blue pair likewise matches snuABE3.1/3.1n. Two independent elements (heat-map rows and Results text) name 4.1/4.1n against the key's 4.2/4.2n, so the key label is the odd one out. No separate snuABE4.2 experiment at the VHL locus is described anywhere.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Fig. 5c, PCSK9 exon 1 panel

  62

• p. 9 Fig. 5c, PCSK9 exon 1 panel (right-hand number over the last residue)

  69

• p. 9 Fig. 5c, PCSK9 exon 4 panel

  212

• p. 9 Fig. 5c, PCSK9 exon 4 panel (right-hand number over the last residue)

  219

Only the authors can say what is right here.

Judge's note: Viewed the full page main-p009 and crop C028-b, which renders Fig. 5c clearly: each panel prints exactly six amino-acid letters over six purple codon bars before the red splice-donor base, with the left number above the first residue and the right number above the last, the same convention used in Fig. 5a (crop C026-a: "101" over L and "113" over R for the thirteen residues L P P G T G R R I H N Y R, and the 332G>A/S111N annotation confirms the eleventh letter is residue 111). The layout text layer independently shows "62 ... 69" and "212 ... 219" above "F H R C A K" and "T R F H R Q". Six residues beginning at 62 end at 67 and beginning at 212 end at 217, so the two panel labels are internally inconsistent by two positions each. Which end of each pair is wrong cannot be settled from the files (the true PCSK9 residue numbering is not in the article), hence author query.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 12 Methods, Amplicon sequencing and data analysis

  https://github.com/BaeLab/ABE_detector_SUB_detector

• p. 14 Code availability

  The analysis code used in this study is available from GitHub (https://github.com/BaeLab/ABE-USB_detector)

• p. 14 Methods references, ref. 61

  Im, H. W. et al. BaeLab/ABE-USB_detector: ABE-USB_detector.

• p. 14 Code availability

  https://github.com/BaeLab/ABE-USB_detector

Only the authors can say what is right here.

Judge's note: Read crop C029-a (Methods, Amplicon sequencing: "Analysis of A-to-I editing efficiency at target sites was performed using the ABE_detector pipeline and total base conversion rates were calculated using the allsub_detector pipeline (https://github.com/BaeLab/ABE_detector_SUB_detector)"), crop C029-b (Code availability: "The analysis code used in this study is available from GitHub (https://github.com/BaeLab/ABE-USB_detector) and archived at Zenodo (https://doi.org/10.5281/zenodo.20656665)61") and crop C029-c (ref. 61, Zenodo record named ABE-USB_detector). The Code availability statement covers "the analysis code used in this study", i.e. the same ABE_detector/allsub_detector pipeline described in the Methods, and only one repository is archived; the two URLs cannot both be right, but only the authors can say which.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 21 Extended Data Fig. 7 caption

  at representative endogenous loci in HEK293T and HEK293 cells. The bar graphs show the mean ± s.d. with error bars (n = 3 independent biological replicates). All experiments were conducted in HEK293T cells.

• p. 21 Extended Data Fig. 7d, panel titles

  HEK293

• p. 21 Extended Data Fig. 7 legend, last sentence

  All experiments were conducted in HEK293T cells.

• p. 21 Extended Data Fig. 7 legend, item c,d

  at representative endogenous loci in HEK293T and HEK293 cells

Judge's note: Read the full page image p021. The legend reads: 'c,d, Evaluation of ngRNA positioning effects on editing efficiency at representative endogenous loci in HEK293T and HEK293 cells. The bar graphs show the mean ± s.d. with error bars (n = 3 independent biological replicates). All experiments were conducted in HEK293T cells.' In the figure the four panels of c are titled 'HEK293T' (HEK3, FANCF, AAVS1, RNF2) and the four panels of d are titled 'HEK293' (HEK3, FANCF, AAVS1, RNF2), so the two cell lines are deliberately distinguished and the unrestricted closing sentence (boilerplate repeated in other Extended Data legends, e.g. Fig. 1 and Fig. 2) contradicts the legend's own panel description and the panel titles. Label/population error; the plotted data are unaffected.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Discussion

  requiring only optimization of the 3′ extension length (typically 24–32 nt)

• p. 4 Results, Characterization of optimized conditions of snuABE3.1

  the highest editing efficiencies were consistently observed with tagRNA lengths between 24 and 33 nt

• p. 2 Fig. 1a schematic, tagRNA extension annotation

  24–33 nt

• p. 4 Results

  tagRNA lengths between 24 and 33 nt

Judge's note: Read crop C026-a/-e/-g (Discussion p. 9, "typically 24-32 nt"), C026-b/-d/-f (Results p. 4, "between 24 and 33 nt (Fig. 3a and Extended Data Fig. 1)", with tagRNAs tested "with 3' extensions ranging from 15 to 69 nt") and C026-c (Fig. 1a schematic on main-p002, annotation "24-33 nt"). The same quantity is printed with two different upper bounds; the figure and Results agree on 33 nt, so the Discussion's 32 nt is the outlier.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Fig. 4 caption, replicate summary

  experiments shown were performed with n = 3 (f,j,n), n = 4 (h) and n = 2 (i,m) independent biological replicates

• p. 6 Fig. 4 caption, panel h description

  Editing efficiencies at each target site were measured from three independent biological replicates

• p. 6 Fig. 4 caption, panel h statistics

  across the five matched target sites (n = 5 paired target sites)

• p. 6 Fig. 4 legend, closing replicate statement

  n = 3 (f,j,n), n = 4 (h) and n = 2 (i,m) independent biological replicates

Only the authors can say what is right here.

Judge's note: Read crops C017-a, C017-b, C017-c, C017-f and the panel itself on main-p007. The legend for h defines three replicates per target site and five paired target sites per cell line, and the paired t-test is declared on n = 5 paired sites; panel h plots roughly five points per cell line over four cell lines (HEK293T, HEK293, HeLa, U2OS). The closing summary nevertheless assigns "n = 4" to h, which matches neither 3 nor 5; it would match the four cell lines, but the sentence explicitly labels the numbers "independent biological replicates". The legend therefore leaves the replicate structure of h ambiguous; only the authors can say what n = 4 denotes.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Results, MLH1-SB and additional nick

  both inhibi- tors greatly enhanced editing efficiencies (Fig. 4h)

• p. 6 Fig. 4 caption, panel h

  h, Fold changes in editing efficiency induced by MLH1-SB, calculated by comparing snuABE4.1-MLH1-SB with snuABE4.1 across five target sites in each cell line

• p. 7 Fig. 4h, plot key

  snuABE4.2

• p. 6 MLH1-SB section

  in MMR-proficient HEK293, HeLa and U2OS cells, both inhibitors greatly enhanced editing efficiencies (Fig. 4h)

Only the authors can say what is right here.

Judge's note: Read crops C018-a/-d/-f (Results sentence, p. 6), C018-b (Fig. 4 legend h: "Fold changes in editing efficiency induced by MLH1-SB, calculated by comparing snuABE4.1-MLH1-SB with snuABE4.1 across five target sites in each cell line"; "Exact P values were as follows: HEK293T, P = 0.2944; HEK293, P = 0.0103; HeLa, P = 0.0417; U2OS, P = 0.0783. Asterisks indicate comparisons with P < 0.05"), and the panel on main-p007 at full page: the only key entry in h is "snuABE4.2", the y axis is "Fold change compared to snuABE4.1", and the four cell-line columns are annotated NS, *, *, and "P = 0.078". So the cited panel contains no MLH1dn data (those are in Supplementary Fig. 2, cited two sentences earlier) and, by the paper's own P < 0.05 criterion, U2OS is not flagged. The cited panel does not support "both inhibitors" nor "greatly enhanced" for all three cell lines as described.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send