AI-Guided Redesign of Laboratory-Evolved Reverse Transcriptases Enhances Prime Editing

• p. 5 Results, pooled ClinVar screen discussion

  Tf1 retrotransposon-derived RT that naturally evolved at 32 degC (ref. 65)

• p. 9 Results, eVLP paragraph

  whereas wild-type Tf1 RT from S. pombe exhibits an activity optimum at 32 degC (ref. 65)

• p. 13 Reference list, ref. 65

  Behrens, R. Fission yeast retrotransposon Tf1 integration is targeted to 5' ends of open reading frames. Nucleic Acids Res. 28, 4709-4716 (2000).

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: whether the Behrens 2000 Nucleic Acids Research paper on Tf1 integration-site targeting reports a temperature optimum for Tf1 reverse transcriptase activity. I verified from the files only what is printed: the p013 page image shows reference 65 exactly as quoted, an integration-site mapping study; the p009 page image (crop C001-c) shows 'whereas wild-type Tf1 RT from S. pombe exhibits an activity optimum at 32 °C (ref. 65)'; crop C009-a shows the p005 use '...the Schizosaccharomyces pombe Tf1 retrotransposon-derived RT that naturally evolved at 32 °C (ref. 65)'. The cited title concerns where Tf1 integrates, not RT enzymology, so the support for the second, biochemical claim is doubtful, but the content of the cited paper is not in the files and cannot be checked here. Rated S2 as a citation-support issue.

Second check: another model agreed

• p. 12 Reference list, ref. 14

  Gori, J. L. et al. Prime editing for p47phox -deficient chronic granulomatous disease. N. Engl. J. Med. 394, 1195-1203 (2025).

• p. 12 Reference list, ref. 59

  Musunuru, K. et al. Patient-specific in vivo gene editing to treat a rare genetic disease. N. Engl. J. Med. 392, 2235-2243 (2025).

• p. 12 Reference list, ref. 60

  Longhurst, H. J. et al. CRISPR-Cas9 in vivo gene editing of KLKB1 for hereditary angioedema. N. Engl. J. Med. 390, 432-441 (2024).

• p. 12 Reference list, ref. 61

  Gillmore, J. D. et al. CRISPR-Cas9 in vivo gene editing for transthyretin amyloidosis. N. Engl. J. Med. 385, 493-502 (2021).

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: the New England Journal of Medicine issues two volumes per calendar year. Read the reference list on the page image main-p012.png (crops C026-a to C026-d). The other NEJM entries in the same list are 385 (2021), 390 (2024) and 392 (2025); on a two-volumes-per-year scale those are consistent with each other (2021 = 384-385, 2024 = 390-391, 2025 = 392-393) and place volume 394 in 2026. The paper itself was accepted 27 April 2026 and published 21 May 2026 (p.1), so a 2026 citation is feasible. I can confirm only what the page prints and the internal inconsistency of the volume-year pattern; the authors should check the year.

Second check: another model agreed

• p. 1 Abstract

  up to twofold higher intracellular prime editor protein levels following mRNA delivery

• p. 7 Results, 'Redesign improves expression level and thermostability of RTs'

  PE8max variants at 8 h after transfection increased by 1.2-, 2.3-, 2.0- and 2.1-fold, respectively, compared to their parental counterparts

• p. 9 Results, in vivo protein expression

  redesigned variants PE8c, PE8d and PE8max exhibited 1.7-, 2.4- and 1.8-fold higher peak protein levels in vivo

• p. 7 Results, Redesign improves expression level

  increased by 1.2‑, 2.3‑, 2.0‑ and 2.1‑fold

Only the authors can say what is right here.

Judge's note: Read the abstract on main p001 (crop C001-a and full page): the wording is "up to twofold higher intracellular prime editor protein levels following mRNA delivery". Read p007 (crop C001-b): anti-Cas9 ELISA on Hepa1-6 lysates gives peak protein increases of 1.2-, 2.3-, 2.0- and 2.1-fold at 8 h (Fig. 4b). Read p009 (crop C001-c): in vivo peak protein levels 1.7-, 2.4- and 1.8-fold (Fig. 5g). The abstract's stated ceiling (twofold) is below the maxima it summarises (2.3 in cells, 2.4 in liver), both from mRNA/LNP delivery and the same anti-Cas9 ELISA readout. I searched the article for a restriction that would make 'twofold' a different population (e.g. a single variant or a single time point) and found none. Rated S3 because this is a wording/ceiling mismatch in a summary rather than a wrong value in the data; only the authors can say whether 'up to twofold' was intended as conservative rounding.

Second check: another model agreed

• p. 2 Introduction, final summary paragraph (right column)

  we designate these new prime editor variants PE8max, PE8c and PE8d

• p. 7 Results, 'Redesigned RTs improve correction of pathogenic mutations'

  we designated PE6a-D37 as PE8a, PE6c-D8 as PE8c, PE6d-D34 as PE8d and PEmax-D27 as PE8max

• p. 11 Discussion, second paragraph

  If payload size is restricted, PE8a should be considered due to its shorter coding sequence (5.6 kilobases)

Judge's note: Read p002 (crop C003-a and full page image): the sentence reads "Based on their general superior performance over PE6 and PEmax-La (PE7) variants, we designate these new prime editor variants PE8max, PE8c and PE8d." Read p007 (crop C003-b region, confirmed in crop C001-b/C013-a): four variants are designated, including PE8a. Read p011 (crop C003-c): "If payload size is restricted, PE8a should be considered due to its shorter coding sequence (5.6 kilobases)". Grep of main.txt shows PE8a used at lines 949, 995, 1007, 1178, 1224, 1532, 1542, 1624, 1641 (including Fig. 4a,b and the Fig. 5i decision schematic). The introductory sentence carries a qualifier ('based on their general superior performance over PE6 and PE7'), and PE8a was not benchmarked against PE7, so a deliberate restriction is possible; that partial explanation is why this is CONFIRMED_MINOR/observation rather than a plain error. The preceding sentence nevertheless names all three lineages (Ec48, Tf1 and M-MLV), so the Ec48-derived PE8a is never introduced.

Second check: another model agreed

• p. 2 Introduction, last sentence of final paragraph (continuing to right column)

  Redesigned RTs based on those from Ec48, Tf1 and M-MLV retain robust reverse transcription activity while exhibiting substantially improved thermal stability and expression levels in mammalian cells.

• p. 8 Fig. 4d legend/panel keys

  Evolved PE6d RT (41.5 °C)

• p. 8 Fig. 4d legend/panel keys

  Redesigned PE8d RT (41.5 °C)

• p. 7 Results, 'Redesign improves expression level and thermostability of RTs'

  Although the PE8d RT did not show improved thermodynamic Tm values compared to the PE6d RT

Only the authors can say what is right here.

Judge's note: Verified the introductory sentence verbatim on the p002 page image (spans the bottom of the left column into the top of the right column). Fig. 4d on the p008 page image and crops C004-b/C004-c show the three DSF panels and their legends: WT-Tf1 43.5 °C, Evolved PE6c-Tf1 44.5 °C, Redesigned PE8c-Tf1 52.5 °C; Engineered PEmax dRNase H 42.5 °C, Evolved PE6d 41.5 °C, Redesigned PE8d 41.5 °C; Engineered PEmax 41.5 °C, Redesigned PE8max 43.5 °C. Crop C004-d confirms the Results sentence conceding no Tm improvement for PE8d. There is no DSF panel for an Ec48/PE6a/PE8a RT anywhere (grep of 'Ec48' and of the Extended Data Fig. 6 legend shows only expression, cell-free yield and a temperature-dependent activity assay). The paper does report improved functional thermostability for PE8d at 57 °C (Extended Data Fig. 6d), which partly supports a broad reading of 'thermal stability'; hence CONFIRMED_MINOR/S3 (overstated summary) rather than a wrong number.

Second check: another model agreed

• p. 11 Discussion, first paragraph

  improved efficiencies for the correction of over 700 therapeutically relevant genetic mutations

• p. 6 Fig. 3b, screen overview table

  700 ClinVar reversions

• p. 2 Introduction, final summary paragraph

  correction efficiencies of 700 pathogenic variants in the ClinVar database

• p. 14 Methods, 'Self-targeting lentiviral pool screen'

  we selected n = 700 edits from library ClinVar

Judge's note: Read crop C022-a and page images main-p011.png, main-p006.png (Fig. 3b: "700 ClinVar reversions", "50 Negative controls", "750 Total epegRNAs", "16,800 Total prime edits"), main-p002.png (introduction: "correction efficiencies of 700 pathogenic variants in the ClinVar database") and the Methods on main-p014.png ("we selected n = 700 edits from library ClinVar"); the Results also state "700 unique prime edits". Every other statement gives exactly 700, and the 50 extra library elements are explicitly negative controls (S. aureus scaffold or scramble spacer), not mutations, so "over 700" overstates slightly. Minor wording only; a reader could also take "over 700" to include the handful of additional arrayed therapeutic edits elsewhere in the paper.

Second check: another model agreed

• p. 3 Fig. 1d, lower panel y-axis

  pLDDT

• p. 19 Extended Data Fig. 1a-c, y-axes (normal orientation)

  pLDDT

Judge's note: Opened crop C005-a and the full p003 page image. The Fig. 1d lower panel y-axis is labelled 'pLDDT' with a break at the top and tick labels 80, 85, 90, 95, 100 reading downwards, so PE6a (highest pLDDT) plots near the bottom and PE6c (lowest) near the top. Crop C005-a also shows Fig. 1e, whose pLDDT axis runs 83 at the bottom to 88 at the top; crop C005-b shows Extended Data Fig. 1a-c with pLDDT likewise increasing upward. The tick labels and the plotted data are mutually consistent, so no value is wrong; the defect is that the same quantity is plotted with opposite axis sense within one figure, which can mislead a reader comparing panels. Recorded as a minor presentational inconsistency.

Second check: another model agreed

• p. 6 Fig. 3b, 'Pooled prime edit screen' tabulation

  750 Total epegRNAs x4 Prime editor RTs x3 Variants per RT x2 Biological replicates 16,800 Total prime edits

• p. 5 Results, pooled screen paragraph

  700 unique prime edits that use eight redesigned RTs and four starting RTs, performed in duplicate, in total, representing the outcomes of 16,800 prime editing experiments

• p. 6 Fig. 3b

  750 Total epegRNAs

• p. 6 Fig. 3b

  16,800 Total prime edits

Judge's note: Read the Fig. 3b panel on the full p006 page image: the two counts 700 and 50 are summed under a rule to '750 Total epegRNAs', the three multipliers follow, and a second rule leads to '16,800 Total prime edits'. 750*4*3*2 = 18,000, whereas 700*4*3*2 = 16,800. I then checked the Results text (crop C011-b, p005), which states 'The resulting dataset comprised 700 unique prime edits that use eight redesigned RTs and four starting RTs, performed in duplicate, in total, representing the outcomes of 16,800 prime editing experiments (Fig. 3b)' - so the printed total 16,800 is itself correct and the panel's own labels distinguish 'Total epegRNAs' from 'Total prime edits', the 50 negative controls not being prime edits. The defect is therefore confined to the visual arithmetic chain, which does not run from the subtotal it displays; the figures themselves are right. Hence CONFIRMED_MINOR/S3 rather than a wrong number.

Second check: another model agreed

• p. 8 Fig. 4 caption, panel d

  Tm values were calculated from the second derivatives of the denaturation curves.

• p. 16 Methods, DSF

  Tm values were determined from the peak of the first derivative of fluorescence with respect to temperature (dF / dT).

• p. 16 Methods, DSF

  determined from the peak of the first derivative of fluorescence

• p. 8 Fig. 4 legend, d

  Tm values were calculated from the second derivatives of the denaturation curves

Only the authors can say what is right here.

Judge's note: Read the Fig. 4 legend on the full p008 page image (panel d sentence verbatim: 'DSF to quantify Tm for purified RT protein variants expressed in E. coli. Thermal unfolding was measured by SYPRO Orange fluorescence, and Tm values were calculated from the second derivatives of the denaturation curves.'). Read the Methods DSF paragraph on crop C014-b (p016): 'Raw fluorescence traces were background subtracted and normalized, and Tm values were determined from the peak of the first derivative of fluorescence with respect to temperature (dF / dT).' The two places name different derivatives as the single procedure used. Because the peak of the first derivative coincides with a zero of the second derivative, the numerical Tm values in Fig. 4d need not differ, so this is a descriptive inconsistency rather than a wrong value; only the authors can say which description is correct. Rated S3, below the ERROR rating one reviewer gave.

Second check: another model agreed

• p. 8 Results, therapeutic applications

  ProteinMPNN-redesigned PE8 variants consistently outperformed their parental counterparts

• p. 9 Results

  We also observed comparable or enhanced editing efficiencies for PE8 variants in human-derived fibroblasts

Judge's note: Read crop C015-a (p008, end of the 'Prime editors with redesigned RTs show improved efficacy in therapeutic applications' opening) for the 'consistently outperformed' sentence, and crop C015-b plus the full p009 page image for the 'comparable or enhanced' sentence covering Extended Data Fig. 7a,b. The same page also reports modest gains at the IL2RB T-cell locus (1.2-fold at the high dose, 1.6-fold at the low dose; Fig. 5c). 'Comparable' outcomes are not outperformance, so the summary wording overstates the subsequent description. I found no definition in the article restricting 'consistently' to a subset of the cell types. The underlying data and figure pointers are sound, so this is a wording/overstatement point only.

Second check: another model agreed

• p. 9 Results, off-target paragraph

  target DNA specificity compared to PE.

• p. 9

  sacrificing product purity or target DNA specificity compared to PE.

Only the authors can say what is right here.

Judge's note: Read crops C017-a/b and full page main-p009.png. The sentence ends "compared to PE." while the two preceding sentences in the same paragraph name the comparators explicitly ("compared to their PE6 or PEmax predecessors"). "PE" is used elsewhere in the paper only as a generic prefix (PE variants, PE-LNP, PE-eVLP, PE protein), never as a named editor, so as printed the comparator is ambiguous; only the authors can say whether a digit or phrase was dropped. Rated S3 (unresolved ambiguity) rather than an outright error.

Second check: another model agreed

• p. 9 Results, HSPCs

  here we achieved 50% editing efficiency with a lower dose (1 µg) of PE8c mRNA

• p. 10 Fig. 5b

  HBB +5 G>A

Only the authors can say what is right here.

Judge's note: Read crop C018-a (text), crop C018-b and full page main-p010.png (Fig. 5b). On the 0-20-40-60 axis of Fig. 5b the PE8c bar top sits clearly below the 50 level (about 45-47%), with its three replicate dots straddling 50 (two near 50, one near 41); the PE6c bar is about 30%, and the same paragraph's "1.6-fold improvement over PE6c" implies about 48%. Methods ("Electroporation of human primary CD34+ HSPCs") confirm Fig. 5b is the high dose (1 µg of editor mRNA) and Extended Data Fig. 7c the low dose (PE8c ~35%), so no other panel supports a 50% value. The text states 50% without an approximation sign; the figure mean reads lower. Source data for Fig. 5b would settle the exact mean, so this is raised as an author query rather than a plain arithmetic error.

Second check: another model agreed

• p. 15 Methods, Electroporation of human primary CD34+ HSPCs

  either a high dose (1 ug of editor mRNA, 180 pmol of epegRNA and 120 pmol of ngRNA) or a low dose (0.5 ug of editor mRNA, 180 pmol of epegRNA and 120 pmol of ngRNA)

• p. 15 Methods, Electroporation of primary human T cells

  either a high dose (1 ug of editor mRNA, 240 pmol of epegRNA and 160 pmol of ngRNA) or a low dose (0.5 ug of editor mRNA, 120 pmol of epegRNA and 80 pmol of ngRNA)

Only the authors can say what is right here.

Judge's note: Read p. 15 image. The HSPC paragraph prints 180 pmol epegRNA and 120 pmol ngRNA in both the high- and low-dose parentheses, differing only in mRNA (1 µg vs 0.5 µg). The immediately preceding T-cell paragraph prints 240→120 pmol epegRNA and 160→80 pmol ngRNA, and the fibroblast paragraph likewise halves mRNA and both guides together. I found no statement anywhere in the Methods or in the Extended Data Fig. 7c legend ('low dose') defining 'dose' as referring to mRNA only. The text is not self-contradictory — the authors may deliberately have varied only the mRNA — so this is an unresolved ambiguity for the authors rather than a demonstrated error; downgraded accordingly.

Second check: another model agreed

• p. 7 Results, arrayed validation of ClinVar variants

  Anderson-Tawil syndrome, molybdenum cofactor deficiency

• p. 6 Fig. 3e, gene label

  KCNJ2

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that the eponym for the KCNJ2 channelopathy is 'Andersen-Tawil syndrome' (after Ellen Damgaard Andersen), not 'Anderson-Tawil'. From the files I confirmed what is printed: crop C001-b (p007) shows the list '...correction of pathogenic variants in MSH6, KIF7, LDLR, CYP4F22, KCNJ2, MOCS1, TRIOBP and MYH14 (Fig. 3e) associated with Lynch syndrome, acrocallosal syndrome, hypercholesterolemia, lamellar ichthyosis, Anderson-Tawil syndrome, molybdenum cofactor deficiency...'. Crop C012-b confirms Fig. 3e pairs KCNJ2 with c.899A>G, and 'respectively' maps KCNJ2 to the fifth disorder, so the spelling applies to the KCNJ2 entry. Purely a spelling slip; meaning unaffected.

Second check: none (judge only)

• p. 22 Extended Data Fig. 4 caption

  across all targets with >0.5% editing

• p. 6 Fig. 3d caption

  elements were filtered to require >=1,000 reads and >=0.5% efficiency

• p. 16 Methods, HTS and analysis

  elements that showed undetectable editing (<0.5% editing)

Judge's note: Read the C036-a crop (Extended Data Fig. 4 legend, right column: 'in prime editing efficiency across all targets with >0.5% editing'), the p. 6 page image (Fig. 3 legend, panel d: 'To apply minimum quality thresholds, elements were filtered to require ≥1,000 reads and ≥0.5% efficiency') and the C036-c crop (Methods: 'elements that showed undetectable editing (<0.5% editing)'). Methods and Fig. 3d agree (keep ≥0.5%); Extended Data Fig. 4 writes the strict inequality for what is evidently the same filter on the same pooled data. The boundary case is of negligible practical consequence, so cosmetic only; either inequality could be the authors' loose shorthand, hence observation.

Second check: none (judge only)

• p. 7

  PE8a, PE8c, PE8d and PE8max showed 7.4, 23, 16 and 24.0% sequence divergence

Judge's note: Opened crop C013-a (p007) and the full p008 page image. The Fig. 4a table row 'Sequence divergence:' reads 7.4%, 23.2%, 16.3%, 24% and the row 'Amino acid changes:' reads 30, 117, 81, 163, which matches the text's amino-acid counts exactly. The text rounds 23.2 to 23 and 16.3 to 16 (one fewer significant figure) but writes the whole number 24 as '24.0', i.e. with a decimal place the figure does not give, inside a single list. No value is wrong; this is a formatting inconsistency only, hence S4.

Second check: none (judge only)

• p. 10 Fig. 5i, right-hand decision box

  If higher efficency is desired, also test PE8d and PE8max (site-dependent effects)

• p. 10 Fig. 5i, bullet list below

  If higher efficiency needed, also test PE8d and PE8max

• p. 10 Fig. 5i, LNP/eVLP branch box

  If higher efficency is desired, also test PE8d and PE8max

• p. 10 Fig. 5i

  If higher efficency is desired,

Judge's note: Read the figure on the page image main-p010.png and cross-checked the PDF text layer (extract/main.layout.txt lines 1003 and 1010), which is a genuine text layer for this page, not OCR. The eVLP branch box reads "If higher efficency is desired," while the bullet list a few lines below in the same panel reads "If higher efficiency needed, also test PE8d and PE8max", so the misspelling is in the published graphic and is internally inconsistent with the same panel.

Second check: none (judge only)

• p. 27 Extended Data Fig. 7a-d, plot legends

  Editng

• p. 10 Fig. 5a legend (correct form for comparison)

  Editing

Judge's note: Opened the full page image main-p027.png: in panels a, b, c and d the colour key above the grey "Indels" entry reads "Editng" in every case, while the running axis label on the same panels is "% Editing or indels" and the Fig. 5a-c keys (crop C046-b / main-p010.png) read "Editing". The string is not in the page text layer (figure text on p. 27 is not extracted), so this was judged from the page image, not from OCR. Consistent misspelling across all four keys; purely cosmetic.

Second check: none (judge only)

• p. 11 Discussion, second paragraph, first sentence

  Our recommendations for selecting PE8 variant depend on editor size requirements and delivery context.

Judge's note: Read crop C023-a and full page main-p011.png. The sentence as printed reads "for selecting PE8 variant depend on"; a determiner ("a"/"the") or the plural "variants" is missing. Purely a dropped word with an evident fix.

Second check: none (judge only)

• p. 14 Methods, Cell culture

  HuH-7 cells were purchased from Cytion (300156)

• p. 15 Methods, In vitro transfection of LNPs

  Hepa1-6 or Huh-7 cells were seeded in 48-well plates

• p. 21 Extended Data Fig. 3c, panel titles

  Top 4 hits, Huh7 cells

Judge's note: Read crops C029-a, C029-b and C029-c plus the page images main-p014.png, main-p015.png and main-p021.png; all three spellings are present as quoted (the Extended Data Fig. 3c panel titles read "Top 4 hits, Huh7 cells / LNP transfection"). Purely a naming-consistency point - HuH-7 is the depositor form and Huh-7 the common form, both defensible - so cosmetic.

Second check: none (judge only)

• p. 15 Methods, LNP production

  (ammonium salt; C14-PEG2000; Avanti Research, 880150) at a molar ratio of OF-02:DOPE:cholesterol:C14-PEG200 = 35:16:46.5:2.5

• p. 15

  OF-02:DOPE:cholesterol:C14-PE G200 = 35:16:46.5:2.5

Judge's note: Read crops C030-a and C030-b and the page image main-p015.png. The full chemical name given is 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000], abbreviated C14-PEG2000, but the molar-ratio expression (which breaks across the line as "C14-PE" / "G200 = 35:16:46.5:2.5") omits the final zero, giving C14-PEG200. The molar ratio itself is internally consistent (sums to 100). A one-character inconsistency in an abbreviation; cosmetic.

Second check: none (judge only)

• p. 16 Methods, Temperature-dependent reverse transcription activity assay

  combining 1 ul of Millenium RNA Markers (1 ug ul-1; Thermo Fisher Scientific, AM7150)

• p. 26 Extended Data Fig. 6 caption, panel d

  Millenium RNA markers (1 ug) were annealed to an oligo(dT) primer

• p. 16 Methods, RT activity assay

  Millenium RNA Markers

• p. 26 Extended Data Fig. 6 legend d

  Millenium RNA markers

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: the Thermo Fisher/Ambion catalogue AM7150 product is named 'Millennium' RNA Markers (double n). What I can verify in the files: the p. 16 page image shows the hyphenated 'Mil-lenium RNA Markers (1 µg µl−1; Thermo Fisher Scientific, AM7150)' and the p. 26 crop/page image shows 'Millenium RNA markers (1 µg)' — both with a single n, so this is not a text-extraction artefact. Cosmetic spelling only; the catalogue number is given and is unambiguous.

Second check: none (judge only)

• p. 23 Extended Data Fig. 5c schematic

  PE8: PE8c/PE8d/PE8max + epegRNA (eSBRMV1 motif)

• p. 24 Extended Data Fig. 5 caption, panel c

  PE8 variants paired with synthetic epegRNA (eSBRNV1) and PE7 paired with synthetic pegRNA with La-accessible end

• p. 24 Extended Data Fig. 5 legend, c

  synthetic epegRNA (eSBRNV1)

• p. 23 Extended Data Fig. 5c schematic

  epegRNA (eSBRMV1 motif)

Judge's note: Read the C037-d and C042-a crops of the p. 23 schematic, which plainly render 'eSBRMV1 motif' under the panel c heading, and the C037-b/C043-h crops of the p. 24 legend, which plainly render '(eSBRNV1)'. Both describe the same arm of the same panel c experiment (PE8 variants with synthetic epegRNA versus PE7 with La-accessible pegRNA), and no second motif is introduced anywhere in the figure or legend, so one spelling is a typographical slip. Name-level typo only, data unaffected.

Second check: none (judge only)

• p. 23 Extended Data Fig. 5e, x-axis label

  Arrayed editing by mRNA at synthetic targert site in Huh-7 cells (%)

• p. 23 Extended Data Fig. 5 c–f axis labels

  synthetic targert site

Judge's note: Read the C039-a and C042-a crops of p. 23 and the full page image: the shared x-axis title under panel e and under panel f both read 'synthetic targert site', and the panel d y-axis title likewise reads 'synthetic targert site'. The corresponding axis in Fig. 3e on p. 6 reads 'Arrayed editing by mRNA at synthetic target site in Hepa1-6 cells (%)', confirming the intended word. Misspelling in rendered figure artwork, not an OCR artefact; cosmetic.

Second check: none (judge only)

• p. 26 Extended Data Fig. 6 legend d

  SuperScript IV at 4.3 uM

Judge's note: Read the C045-a crop and the p. 26 page image: the legend prints '4.3 uM' while the same legend uses the correct glyph elsewhere ('15 µL', '5 mL', '1 µg'), and the corresponding Methods sentence on p. 16 prints 'at 4.3 µM'. So the glyph substitution is on the page, not an extraction artefact. The unit itself is right and the value matches the Methods; typographic only.

Second check: none (judge only)

• p. 30 Reporting Summary, Software and code, Data collection

  Data for DSF was collected on Bio-Rad CFX Maestor software (v2.3).

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that Bio-Rad's CFX analysis package is named "CFX Maestro", which is not stated anywhere in the files. What I can verify from the files: the page image main-p030.png shows the Data collection box reading "Data for DSF was collected on Bio-Rad CFX Maestor software (v2.3)", and the same spelling appears in the extracted layout text (main.layout.txt line 2026), so it is not an OCR artefact. Only a spelling of a product name; cosmetic.

Second check: none (judge only)

• p. 33 Reporting Summary, Eukaryotic cell lines, Cell line source(s)

  Gesicle 293T cells (632617) were obtain from Takara.

Judge's note: Read the full page image main-p033.png; the Cell line source(s) box reads "...Gesicle 293T cells (632617) were obtain from Takara. Human CD34+ hematopoietic stem and progenitor cells (HSPCs) were obtained from unidentified healthy adult donors...", with the correct "were obtained" used in the adjacent sentences. The same wording appears in the extracted layout text (line 2190), so it is not an OCR error. Purely a missing word ending.

Second check: none (judge only)