A Draft Human Pangenome Reference
2023 · paper · Biomedicine
Presents a draft human pangenome reference built from 47 phased, diploid assemblies of genetically diverse individuals. Compared with GRCh38, it adds 119 million base pairs, reduces small-variant errors by 34% and raises structural variants detected per haplotype by 104%.
Checked: PDF from europepmc.org, fetched 4 Oct 2026. Page numbers are those of that PDF.
12 findings
Substance "Only repeats with a period size between 6,000 and 10,000 bp, total length >100 bp ... were selected, leaving a total number of 98,021" cannot be right: the TRF command in the same sentence ("trf hg38.fa 2 7 7 80 10 50 500") caps the period at 500 bp, and the article defines a VNTR as unit motif length ≥7 bp.
- p. 30 Methods, Simulating and mapping VNTR reads
Only repeats with a period size between 6,000 and 10,000 bp, total length >100 bp
- p. 30 Methods, Simulating and mapping VNTR reads
trf hg38.fa 2 7 7 80 10 50 500 -f -d
- p. 6 Fig. 3 legend
a tandem repeat with the unit motif length ≥7 bp
- p. 30 Methods, Simulating and mapping VNTR reads
period size between 6,000 and 10,000 bp
Only the authors can say what is right here.
Judge's note: Read in the page 30 crop; the range is printed as "6,000 and 10,000 bp". Three things in the paper contradict it: the quoted TRF command ends in 500 (maximum period size), a total-length filter of >100 bp would be vacuous for units of 6 kb or more, and the Fig. 3 legend (page 6 crop) defines VNTR as motif ≥7 bp and STR as ≤6 bp. A lower bound of 6 bp is the likely intent, but the upper bound of 10,000 also conflicts with the 500 limit, so the authors must say what the range was. No explanatory statement found elsewhere in the article text.
Second check: another model agreed
Report a problem with this finding
Substance The filtered PanGenie SV set is described in the text as "28,434 deletion alleles, 84,752 insertion alleles and 26,439 other SV alleles" and in the Fig. 6 legend as "(28,433 deletions, 84,755 insertions, 32,431 other alleles)"; all three counts differ, the 'other' class by 5,992.
- p. 10 SV genotyping, second paragraph
28,434 deletion alleles, 84,752 insertion alleles and 26,439 other SV
- p. 9 Fig. 6 legend, panel c
(28,433 deletions, 84,755 insertions, 32,431 other alleles)
- p. 10 SV genotyping, left column
28,434 deletion alleles, 84,752 insertion alleles and 26,439 other SV alleles
- p. 9 Fig. 6c legend
28,433 deletions, 84,755 insertions, 32,431 other alleles
Only the authors can say what is right here.
Judge's note: Both passages read on the page images (page 10 left column; page 9 Fig. 6 legend). The legend says 'The PanGenie genotypes include all variants contained in the filtered set', the same set the text calls the 'filtered, high-quality subset', so the counts should agree. I searched the Methods: they give only shares (50%, 33% and 26% of deletion, insertion and other alleles in the filtered set) and point to Supplementary Table 18, which is not among the files. The post-merge count n = 100,442 in panel d is a separate, labelled quantity and matches neither. A subset restriction could not explain the legend's larger insertion and 'other' counts. Which set of numbers is right cannot be settled without Supplementary Table 18.
Second check: another model agreed
Report a problem with this finding
Substance "with 6.29% fewer SNPs (previous N = 3,735,605, pruned N = 3,676,746)": the two counts give a reduction of 1.58%, not 6.29%.
- p. 32 Methods, Population genetic analyses, third paragraph
with 6.29% fewer SNPs (previous N = 3,735,605, pruned N = 3,676,746)
- p. 32 Population genetic analyses, paragraph 'Genome-wide, we found that pruning...'
6.29% fewer SNPs (previous N = 3,735,605, pruned N = 3,676,746)
Only the authors can say what is right here.
Judge's note: Numbers read in the page 32 crop. The genome-wide figure in the preceding sentence (1.188% from 23,272,652 to 22,996,113) reproduces exactly, so the calculation method is not in doubt. Neither denominator gives 6.29% for the acrocentrics (1.58% of the previous count, 1.60% of the pruned count). I read the surrounding Methods text and found no statement that the percentage refers to a different region (such as p-arms only) from the counts. The qualitative claim that the acrocentric reduction is higher than genome-wide still holds at 1.58%. Either the percentage or one of the counts is wrong; only the authors can say which.
Second check: another model agreed
Report a problem with this finding
Reference or label Fig. 2d right-hand axis "Number of samples with additional copies" (legend: "Grey bars, the number of samples with additional copies") runs to 90 and the tallest grey bars reach about 88, more than the 47 samples; the bars appear to count haploid assemblies.
- p. 4 Fig. 2d, right-hand axis and legend
Number of samples with additional copies
- p. 4 Fig. 2 legend, panel d
Grey bars, the number of samples with additional copies
Only the authors can say what is right here.
Judge's note: Viewed the Fig. 2 crop. The right axis of panel d is ticked 0 to 90 and its 90 aligns with 25 on the left axis; the first grey bar (ZNF595) reaches about 24.5 on the left scale, about 88 on the right. In panel e the grey bars likewise belong to the right axis. There are 47 samples (94 haploid assemblies), and the text counts CNV frequency 'in 10% or more of haploid assemblies', so the unit is most likely haplotypes and the label 'samples' is wrong. Bar heights are read from the image and are approximate; the authors should confirm the unit.
Second check: another model agreed
Report a problem with this finding
Reference or label "the two-pass mode of DeepVariant (v.1.1.0)107 with WhatsHap (v.1.1)108" cites ref. 107, which is the Winnowmap2 paper (Jain et al.); DeepVariant is ref. 40 (Poplin et al.) elsewhere in the article.
- p. 25 Methods, Calling variants from PacBio HiFi reads
the two-pass mode of DeepVariant (v.1.1.0)107 with
- p. 33 Reference list, entry 107
Long-read mapping to repetitive reference sequences using Winnowmap2
- p. 25 Variant benchmarking, Calling variants from PacBio HiFi reads
two-pass mode of DeepVariant (v.1.1.0)107 with WhatsHap (v.1.1)108
- p. 33 Reference list, ref. 107
Jain, C., Rhie, A., Hansen, N. F., Koren, S. & Phillippy, A. M. Long-read mapping to repetitive reference sequences using Winnowmap2.
Judge's note: Page 25 crop shows superscript 107 on both "Winnowmap2 (v.2.03)" and, one paragraph later, "DeepVariant (v.1.1.0)". The reference list on page 33 gives 107 as Jain et al., 'Long-read mapping to repetitive reference sequences using Winnowmap2'. Ref. 40 on page 12 is Poplin et al., the DeepVariant paper, and the main text cites DeepVariant as ref. 40 three times. The 107 on DeepVariant is a carried-over citation number.
Second check: another model agreed
Report a problem with this finding
Reference or label Extended Data Fig. 5 groups are labelled "% Protein-coding genes mapped" and "% Noncoding genes mapped" and the legend says "show the percentage of protein-coding genes", but the y-axis runs 0.95-0.99, i.e. fractions, not percentages.
- p. 39 Extended Data Fig. 5, x-axis group labels and y-axis
% Protein-coding genes mapped
- p. 39 Extended Data Fig. 5 legend
show the percentage of protein-coding genes
- p. 19 Extended Data Fig. 5 y-axis ticks
0.99
Judge's note: Viewed the page 39 image. The y-axis ticks are 0.95, 0.96, 0.97, 0.98, 0.99 and the axis has no title. The group labels beneath carry a '%' sign and the legend says 'percentage'. The plotted values are proportions, so the unit label is wrong; the meaning is clear from context. Fix: multiply the axis by 100 or label it as a proportion.
Second check: another model agreed
Report a problem with this finding
Reference or label Allele-filtered graph is said to be produced "using vg clip -d 9 -m 10000" (PDF page 23) but is identified as "clip.d9.m1000.D10M.m1000" (PDF page 27): -m 10000 against m1000.
- p. 23 Methods, MC, step (5)
-d 9 -m 10000
- p. 27 Methods, Alignment of reads to the pangenome
clip.d9.m1000.D10M.m1000
Only the authors can say what is right here.
Judge's note: Both crops read as quoted. The page 27 passage says it refers to the allele-filtered graph 'as defined in the MC subsection', so both describe the same graph. In the label, d9 matches -d 9 and D10M matches -D 10000000, so m1000 should match -m; a search of the text for 'm1000' and 'vg clip' found no other value or explanation. One of the two is wrong, and only the authors can say which.
Second check: another model agreed · checked by hand
Report a problem with this finding
Reference or label Extended Data Fig. 10b legend says "Mapping rate and false discovery rate are stratified by mapping quality producing the curves shown. The MC graph is compared to a graph derived from the 1KG variant calls and to GRCh38." but panel b shows single points of mapping precision against mapping recall for only two references ("MC spliced graph", "Spliced GRCh38 reference").
- p. 44 Extended Data Fig. 10 legend, panel b
Mapping rate and false discovery rate are stratified by mapping quality producing the curves shown.
- p. 44 Extended Data Fig. 10 legend, panel b
The MC graph is compared to a graph derived from the 1KG variant calls and to GRCh38.
- p. 44 Panel b axes and key
Mapping precision
- p. 44 Extended Data Fig. 10 caption, b
The MC graph is compared to a graph derived from the 1KG variant calls and to GRCh38
Only the authors can say what is right here.
Judge's note: Viewed the page 44 image. Panel b ('Simulated reads (HG002)') plots mapping precision (y) against mapping recall (x) with one marker per method (vg mpmap unique/multi, STAR unique/multi). There are no curves, no mapping-quality stratification, and the Reference key lists only the MC spliced graph and the spliced GRCh38 reference; no 1KG-derived graph appears. The article text has no statement that reconciles this. The legend describes a different version of the panel; the authors need to say which is right.
Second check: another model agreed
Report a problem with this finding
Wording CNV gene counts do not reconcile: "1,115 protein-coding gene families", "71% of CNV genes appeared in a single haplotype", strata "118 out of 771" singleton, "83 out of 140" low frequency (<10%) and "44 out of 54" high frequency (sum 965), then "58 genes were CNVs in 10% or more of haploid assemblies".
- p. 5 Annotating 47 diverse genomes, CNV paragraph
(44 out of 54) of high frequency. Overall, 58 genes were CNVs in 10% or more of haploid assemblies
- p. 5 Annotating 47 diverse genomes, CNV paragraph
low frequency (<10%) and high frequency
- p. 5 Annotating 47 diverse genomes (continued), left column
There were 1,115 protein-coding gene families
- p. 5 same paragraph
71% of CNV genes appeared in a single haplotype
Only the authors can say what is right here.
Judge's note: Page 5 image reads as quoted. 771 + 140 + 54 = 965, 150 short of 1,115. 771/1,115 = 69.1% and 771/965 = 79.9%, neither is 71%. The high-frequency stratum (not <10%) has 54 genes but the next sentence gives 58 at 10% or more. The within-stratum percentages reproduce (15.3%, 59.3%, 81.5%). I searched the Methods ('Gene duplication analysis', 'Gene family analysis') and the Fig. 2 legend ('out of all 1,115 duplicated genes') and found no statement of a subset for the stratified analysis or of the denominator for 71%. A genes-versus-gene-families or SD-annotatable subset may explain it, but only the authors can say.
Second check: another model agreed
Report a problem with this finding
Wording "the number of SVs per sample within 'easy' genomic regions was consistent across all three call sets, particularly in low-mappability and tandem repeat regions, and the use of our pangenome reference led to substantial gains (Fig. 6d)": the 'particularly in...' clause is attached to consistency, but Fig. 6d shows the largest differences in exactly those regions.
- p. 10 SV genotyping, end of section
consistent across all three call sets, particularly in low-mappability and tandem repeat regions, and the use of our pangenome reference led to substantial gains (Fig. 6d)
- p. 29 Methods, Number of SVs per sample
In the ‘easy’ regions, the number of SVs per sample was consistent across all three call sets. For the other regions, however
- p. 9 Fig. 6d
Low mappability
- p. 9 Fig. 6d, 'Low mappability and SegDup' and 'Tandem repeat' groups
Tandem repeat
Judge's note: Page 10 image for the sentence, page 9 image for Fig. 6d, zoomed page 29 for the Methods. Fig. 6d: the three call sets overlap in 'Easy' (about 3,500 each); in 'Low mappability and SegDup' HPRC is about 6,000 against about 4,000 and 3,000; in 'Tandem repeat' about 11,000 against about 6,500 and 2,500. Methods (page 29) state the intended meaning: 'In the 'easy' regions, the number of SVs per sample was consistent across all three call sets. For the other regions, however, ... more variant alleles ..., especially insertions and variants in regions of low mappability and tandem repeats'. The qualifier belongs with the gains, not the consistency; as printed the sentence is also self-contradictory (easy regions cannot be 'particularly' low-mappability regions).
Second check: another model agreed
Report a problem with this finding
Wording Flagger on T2T-CHM13: "about 12.77 Mb (around 0.41%)" flagged; rDNA arrays, "flagged as falsely duplicated in their entirety" and "about 9.92 Mb in total", were excluded, "which decreased the number of unreliable bases to 5.58 Mb (around 0.18%)"; 12.77 − 9.92 is 2.85 Mb, not 5.58 Mb.
- p. 20 Methods, Assessing T2T-CHM13 using Flagger
we excluded rDNA arrays (about 9.92 Mb in total) from the reference evaluation, which decreased the number of unreliable bases to 5.58 Mb (around 0.18%)
- p. 20 Methods, Assessing T2T-CHM13 using Flagger
about 12.77 Mb (around 0.41%)
- p. 20
we excluded rDNA arrays (about 9.92 Mb in total) from the reference evaluation, which decreased the number of unreliable bases to 5.58 Mb
Only the authors can say what is right here.
Judge's note: Read on the crop of PDF page 20. The drop from 12.77 to 5.58 Mb is 7.19 Mb, not 9.92 Mb, although the text says the rDNA arrays were flagged in their entirety. The following step is consistent (5.58 − 2.76 = 2.82 Mb, printed 2.82 Mb), and the percentages fit a roughly 3.1 Gb genome. I searched the text for rDNA: nothing says only part of the arrays was flagged or gives another rDNA size. Which figure is wrong (9.92, 5.58 or 'in their entirety') only the authors can say. The headline 0.09% is unaffected, so I rate it S3 rather than S1.
Second check: another model agreed
Report a problem with this finding
Wording "genotype concordances of all lenient variants (brown, dark purple) were slightly higher compared with the concordances we observed for the unfiltered set": the paragraph defines subsets of "our filtered set"; 'lenient' is used elsewhere only for the HGSVC call set.
- p. 30 Methods, Number of SVs per sample, first paragraph on page
ances of all lenient variants (brown, dark purple) were slightly higher
Only the authors can say what is right here.
Judge's note: Read in the page 30 crop. The paragraph lists three subsets (novel, known, all variants in our filtered set) and then compares "all lenient variants" with the unfiltered set, so the filtered set seems meant. Everywhere else in the article 'lenient' names the HGSVC PanGenie v.2.0 set (Fig. 6 legend, Results, Methods), apart from an unrelated 'more lenient second merge' in svtools. The HPRC filtering section defines 'unfiltered', 'positive' and 'filtered' sets, not a lenient one. Supplementary Fig. 29 is not in the files, so its legend labels could not be checked; that would settle whether 'lenient' is the figure's own label.
Second check: another model agreed
Report a problem with this finding
32 minor (typos and the like; most checked by the judge only)
Wording "For each class of unreliable region (unknown, erroneous, duplicated, collapsed and haploid)" lists 'haploid' as unreliable, whereas the Fig. 1 legend says "Regions flagged as haploid are reliable (green)".
- p. 20 Methods, SD reliability
For each class of unreliable region (unknown, erroneous, duplicated, collapsed and haploid)
- p. 3 Fig. 1 legend, panel h
Regions flagged as haploid are reliable (green)
Judge's note: The SD-reliability paragraph on PDF page 20 (visible in the C040 crop) and the Fig. 1 legend crop on PDF page 3 both read as quoted. Haploid is the reliable Flagger class by the paper's own definition; the authors evidently meant 'each Flagger class'. Loose wording, meaning recoverable.
Second check: another model agreed
Report a problem with this finding
Wording "1.8% more identical (98.9 compared with 97.1)": the 1.8 is a difference in percentage points, not a relative 1.8%.
- p. 4 Completeness and CNV, SD paragraph
1.8% more identical (98.9 compared with 97.1)
Judge's note: Page 4 image reads as quoted. 98.9 - 97.1 = 1.8 percentage points; the relative difference is 1.85%. The value is right when read as points and the two numbers are given in the parenthesis, so the reader is not misled; wording point only.
Second check: another model agreed
Report a problem with this finding
Wording The AF breakdown gives "81% (44 out of 54) of high frequency" (high frequency being >=10%), while the next sentence says "Overall, 58 genes were CNVs in 10% or more of haploid assemblies"; the paper does not say why 54 and 58 differ.
- p. 5 Results, gene duplications
81% (44 out of 54) of high frequency. Overall, 58 genes were CNVs in 10% or more of haploid assemblies
Judge's note: Read on the page 5 image. Low frequency is defined as <10%, so the high-frequency class should be the >=10% class, yet it has 54 genes against 58 in the next sentence. The three AF classes sum to 965 (771+140+54), not the 1,115 CNV gene families stated earlier, so the stratified analysis evidently covers a subset and the 58 may come from the full set. No statement in the text explains this, so it is an unexplained inconsistency rather than a demonstrable error. Percentages check out.
Second check: another model agreed
Report a problem with this finding
Wording Discussion says "β-defensin (three to seven copies, DEFB107A)", whereas Results say "DEFB107A has three to seven additional copies assembled"; the Discussion drops 'additional' although the same sentence writes "one additional copy" for NOTCH2NL.
- p. 11 Discussion, second paragraph
(three to seven copies, DEFB107A)
- p. 5 Annotating 47 diverse genomes (continued), left column
DEFB107A has three to seven additional copies assembled
Judge's note: Page 11 image (Discussion, second paragraph) and page 5 image (Results). The same range is given as additional copies in Results and as copies in the Discussion. The Discussion sentence is about 'additional genic sequences per haplotype compared with GRCh38' and distinguishes '(one additional copy)' for the next item, so a reader could take 'three to seven copies' as a total. It may be shorthand; minor wording inconsistency.
Second check: another model agreed
Report a problem with this finding
Wording "adding 160 μl Circulomics elution buffer (EB) plus 0.02% Triton X-100 (comprising 316.8 μl EB and 3.2 μl 2% Triton X-100)": the stated components total 320 μl, twice the 160 μl said to be added.
- p. 15 Methods, Nanopore DNA extraction
adding 160 μl Circulomics elution buffer (EB) plus 0.02% Triton X-100 (comprising 316.8 μl EB and 3.2 μl 2% Triton X-100)
Judge's note: Zoomed render of page 15 confirms '160 μl' and '316.8 μl EB'; the '3.2 μl 2% Triton X-100' continuation is from the text layer. 316.8 + 3.2 = 320 μl, and 3.2 μl of 2% in 320 μl gives 0.02% as stated, so the recipe is internally right for a 320 μl batch. Most likely the recipe is a prepared batch from which 160 μl is used per tube, but the text says the 160 μl 'comprises' those volumes. Ambiguity in a protocol description, not a demonstrable error.
Second check: another model agreed
Report a problem with this finding
Wording Reporting summary, Peak calling parameters: "We aligned H3K4me1 and H3K27ac binding sequences ... for each of the 30 H3K4me1 and H3K27ac samples" omits ATAC-seq, whereas the Methods say "for each of the 30 H3K4me1, H3K27ac and ATAC-seq samples" and the same summary lists ATAC-seq sequencing depth.
- p. 62 Reporting summary, Peak calling parameters
for each of the 30 H3K4me1 and H3K27ac samples
- p. 62 Reporting summary, Sequencing depth
ATAC-seq; Average of 279 million reads per sample
- p. 31 Methods, ChIP-seq analysis
for each of the 30 H3K4me1, H3K27ac and ATAC-seq samples
Judge's note: Viewed the page 62 image and the page 31 crop. The Methods (ChIP-seq analysis) name three assays for alignment and peak calling. The reporting summary's Sequencing depth box includes 'ATAC-seq; Average of 279 million reads per sample' but its Peak calling box names only the two histone marks. The summary is a ChIP-seq form, so the omission may be deliberate; it is an inconsistency, not a wrong result.
Second check: another model agreed
Report a problem with this finding
Typo Fig. 1c y-axis label reads "Total lengh (×10⁹)"; should be "Total length".
- p. 3 Fig. 1c, y-axis label
Total lengh (×109)
Judge's note: The crop of Fig. 1c shows the axis label 'Total lengh (×10^9)' with the t missing. The superscript is intact in the image.
Second check: none (judge only)
Report a problem with this finding
Typo "overlap with the pericentomeric or centromeric satellites and SDs": "pericentomeric" should be "pericentromeric".
- p. 4 Completeness and CNV, second paragraph on page
pericentomeric or centromeric satellites and SDs
- p. 4 Completeness and CNV, second paragraph
pericentomeric or centromeric satellites
Judge's note: Page 4 image and text layer both read 'pericentomeric'; the previous paragraph on the same page spells 'pericentromeric'.
Second check: none (judge only)
Report a problem with this finding
Typo "(around 10 Mb and 60 Mb in PCGB)": "PCGB" should be "PGGB".
- p. 5 Measuring pangenome variation, last paragraph
and 60 Mb in PCGB)
- p. 5 Constructing/Measuring pangenome variation, last paragraph
around 10 Mb and 60 Mb in PCGB
- p. 4 Completeness and CNV
pericentomeric or centromeric satellites
- p. 5 Measuring pangenome variation, first paragraph
The Minigraph chart, by virtue of being limited
Judge's note: Page 5 image shows 'PCGB' in the last paragraph, two lines after 'around 190 Mb in PGGB' and 'around 105 Mb in PGGB'. No tool called PCGB is defined. The cluster's other two items ('pericentomeric', 'Minigraph chart') are handled in C007 and C014.
Second check: none (judge only)
Report a problem with this finding
Typo "The Minigraph chart, by virtue of being limited to structural variation, is the smallest": "chart" should be "graph".
- p. 5 Measuring pangenome variation, first paragraph
The Minigraph chart, by virtue of being limited to structural
- p. 5
The Minigraph chart, by virtue of being limited
Judge's note: Page 5 image reads 'The Minigraph chart'. The sentence compares node and edge counts and length with 'the base level graphs' and 'the MC graph', so the Minigraph pangenome graph is meant; 'chart' is a wrong word.
Second check: none (judge only)
Report a problem with this finding
Typo "GPRIN2 is a copy-number polymorphic³² based on read depth and has a sequence resolution of one to three additional copies": a noun is missing after "polymorphic" (e.g. "gene" or "locus").
- p. 5 Annotating 47 diverse genomes, CNV paragraph
GPRIN2 is a copy-number polymorphic32 based on read depth
Judge's note: Page 5 image shows 'a copy-number polymorphic' followed directly by the superscript citation 32, with no noun. Not an extraction loss.
Second check: none (judge only)
Report a problem with this finding
Typo Fig. 5b labels the GRCh38 haplotype "RDH;RHCE"; it should read "RHD;RHCE", as in Fig. 5c and the caption.
- p. 8 Fig. 5b, first haplotype (GRCh38)
RDH;RHCE
- p. 8 Fig. 5c, first row
RHD;RHCE
Judge's note: Zoomed render of Fig. 5b/c on page 8: the first panel title in b clearly reads 'RDH;RHCE' above 'GRCh38'. The first row of the Fig. 5c table reads 'RHD;RHCE' (count 43), the legend lists the gene as RHD, and the caption says 'Structural haplotypes of RHD and RHCE'. Letter transposition in a figure label.
Second check: none (judge only)
Report a problem with this finding
Typo "the gene conversion event covering the second exon RHCE, which was observed in about 25% of assembled haplotypes": 'of' is missing ('the second exon of RHCE').
- p. 10 A pangenome variant resource (continued), left column
event covering the second exon RHCE, which was observed
- p. 10 A pangenome variant resource
covering the second exon RHCE
Judge's note: Page 10 image, left column, top paragraph: the printed text reads 'covering the second exon RHCE' with RHCE in italics and no 'of'. Not an extraction loss.
Second check: none (judge only)
Report a problem with this finding
Typo "using a Dovetail Omni-C kit (Dovetail Genomics) with the following modifications as. First, we aliquoted 1 million cells": stray 'as' (or a missing 'follows') at the end of the sentence.
- p. 16 Methods, Dovetail Omni-C
modifications as. First, we aliquoted
Judge's note: Crop of page 16 shows the printed text 'with the following modifications as. First, we aliquoted'. The sentence ends on a dangling 'as'. Present in print, not an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo Stray word 'as' before the full stop: "with the following modifications as. First, we aliquoted 1 million cells".
- p. 16 Dovetail Omni-C
with the following modifications as.
Judge's note: Crop of PDF page 16 shows the sentence exactly as quoted; 'as' is left dangling (probably a leftover from 'as follows').
Second check: none (judge only)
Report a problem with this finding
Typo Stray or misplaced word 'selection': "To do this, selection we aligned the sequence of the mitochondrial DNA".
- p. 17 Methods, Cleaning steps, last cleaning step
To do this, selection we aligned
Judge's note: Crop of PDF page 17 shows the wording as quoted; the sentence is ungrammatical ('To do this, we aligned' or 'To do this selection, we aligned').
Second check: none (judge only)
Report a problem with this finding
Typo Tool and script names misspelled: "k-mers (hap-mers) were generated using merylu hapmer" and "bash hampers.sh maternal.meryl paternal.meryl sample.meryl".
- p. 18 Methods, Assembly QV assessment
bash hampers.sh maternal.meryl paternal.meryl sample.meryl
- p. 18 Methods, Assembly QV assessment
k-mers (hap-mers) were generated using merylu hapmer
- p. 18 Assembly QV assessment
were generated using merylu hapmer
Judge's note: Crop of PDF page 18 shows both 'merylu hapmer' and 'hampers.sh' in the typeset page. The paper's own wording 'hap-mers' and its use of 'meryl' everywhere else point to 'hapmers.sh' and a stray 'u' in 'merylu'. That the Merqury script is really called hapmers.sh is outside knowledge, but the inconsistency with the paper's own terms is visible on the page.
Second check: none (judge only)
Report a problem with this finding
Typo Misspelled tool name: "using Dipcall (v.0.3) (using mimimap2 (v.2.2.4))"; the aligner is minimap2 everywhere else.
- p. 18 Methods, GIAB-based assembly quality assessment
(using mimimap2 (v.2.2.4))
- p. 18
using mimimap2 (v.2.2.4))38
Judge's note: Crop of PDF page 18 shows 'mimimap2' in the page image; the text spells it 'minimap2' elsewhere (for example PDF pages 17 and 21).
Second check: none (judge only)
Report a problem with this finding
Typo "compared the length of identity of SDs that overlapped the different types of errors" should read 'length and identity'.
- p. 20 Methods, SD reliability
compared the length of identity of SDs
Judge's note: Crop of PDF page 20: the same sentence begins 'we calculated the average length and identity of SDs'; 'length of identity' is a slip for 'length and identity'.
Second check: none (judge only)
Report a problem with this finding
Typo "The sequences of these anchors were them mapped against the target genome": 'them' should be 'then'.
- p. 21 Methods, Ensembl mapping pipeline for annotation
were them mapped against the target genome
- p. 21 Ensembl mapping pipeline, anchor points
The sequences of these anchors were them mapped against the target genome
Judge's note: Crop of PDF page 21 shows 'were them mapped' in the page image.
Second check: none (judge only)
Report a problem with this finding
Typo Duplicated words: "at least 80% of the of the same genes in common".
- p. 21 Methods, Ensembl mapping pipeline for annotation
at least 80% of the of
- p. 21 Ensembl mapping pipeline, second approach
at least 80% of the of the same genes in common
Judge's note: The C042 crop of PDF page 21 covers this passage in the right-hand column: one line ends 'of the of' and the next begins 'the same genes', so 'of the' is repeated in the typeset page.
Second check: none (judge only)
Report a problem with this finding
Typo The last contig-length bin is printed without its number: "<75% for [100 kb,1 Mb) and <70% for ≥mb.)"; the preceding bin ends at 1 Mb, so "≥1 Mb" is meant.
- p. 23 Methods, MC, step (3) Chromosome decomposition
<70% for ≥mb.)
- p. 23 MC step (3) Chromosome decomposition
<70% for ≥mb.
Judge's note: Seen in the crop and the full page image of PDF page 23: the text reads "≥mb." with no numeral and a lower-case unit. The series [1,10 kb), [10 kb,100 kb), [100 kb,1 Mb) makes the intended value 1 Mb evident. The threshold is recoverable from context, so this is cosmetic.
Second check: none (judge only)
Report a problem with this finding
Typo "the multiallelic sites were splitted into biallelic records" should read "were split".
- p. 26 Methods, Benchmarking variants
were splitted into biallelic records
Judge's note: Visible in the page 26 crop. The same sentence earlier in the Methods (PDF page 24 text) reads "were split into biallelic records using bcftools norm -m -any".
Second check: none (judge only)
Report a problem with this finding
Typo Missing full stop: "from RTG tools (v.3.12.1)83 The comparison was performed independently for each individual."
- p. 26 Benchmarking variants
RTG tools (v.3.12.1)83 The comparison was performed independently
Judge's note: Visible in the page 26 crop: the superscript 83 is followed directly by the capitalised "The comparison" with no period.
Second check: none (judge only)
Report a problem with this finding
Typo "which has been previously reposed by several studies5,8" should read "previously reported".
- p. 29 Methods, Number of SVs per sample
which has been previously reposed by several studies
- p. 29 Methods, Number of SVs per sample, right column
which has been previously reposed by several studies5,8
Judge's note: Visible in the page 29 crop (right column). 'Reposed' makes no sense here; 'reported' is clearly meant.
Second check: none (judge only)
Report a problem with this finding
Typo Duplicated word: "we used previously published data NA12878 RNA-seq data (SRR1153470)127 and the ENCODE project (ENCSR000AED, replicate 1)".
- p. 31 Methods, RNA-seq mapping evaluation
published data NA12878 RNA-seq data (SRR1153470)
Judge's note: Visible in the page 31 crop. 'data' appears twice around 'NA12878 RNA-seq', and 'and the ENCODE project' lacks 'from'.
Second check: none (judge only)
Report a problem with this finding
Typo Award number "U01HG01973" has a five-digit serial where every other NIH award in the paragraph has six (e.g. "1U01HG010973").
- p. 33 Acknowledgements
U01HG01973
- p. 33 Acknowledgements
1U01HG010973
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that NIH award serial numbers have six digits and that no award U01HG01973 exists. The page 33 crop shows "U01HG01973" in the second funding list, against six-digit serials for all the others (U41HG010972, 1U01HG010973, U01HG010961, 5U01HG010971 and so on). A dropped digit is likely, but the intended number cannot be determined from the paper.
Second check: none (judge only)
Report a problem with this finding
Typo Missing comma between award numbers: "U24HG010262 U01HG010963,".
- p. 33 Acknowledgements
U24HG010262 U01HG010963
Judge's note: Visible in the page 33 crop: every other award number in the list is comma-separated, but these two are separated only by a space.
Second check: none (judge only)
Report a problem with this finding
Typo Extended Data Fig. 5 x-axis label reads "Cactus/MG GRCh38" while the legend calls it "CAT run on the MC graph based on GRCh38"; "MG" is not defined anywhere in the article.
- p. 39 Extended Data Fig. 5 x-axis category labels
Cactus/MG GRCh38
- p. 39 Extended Data Fig. 5 caption
CAT run on the MC graph based on GRCh38
Judge's note: Viewed the page 39 image: both x-axis groups carry the label 'Cactus/MG GRCh38'. A search of the article text finds no other use of 'MG' as an abbreviation; the paper uses 'MC' for Minigraph-Cactus throughout. 'Cactus/MG' is most likely 'Cactus/Minigraph' written the other way round, so it is an undefined, inconsistent label rather than a clear error.
Second check: none (judge only)
Report a problem with this finding
Typo Extended Data Fig. 7a legend: "for each in the 1KG samples" is missing a word (should read 'for each of the 1KG samples' or 'for each sample in the 1KG').
- p. 41 Extended Data Fig. 7a legend
for each in the 1KG samples
Judge's note: Viewed the page 41 image. The legend reads 'Number of variants with at least one alternate allele (i.e. excluding homozygous for the reference allele) for each in the 1KG samples.' The wording is as quoted and ungrammatical.
Second check: none (judge only)
Report a problem with this finding
Typo Extended Data Fig. 11 legend: "The left plot excludes the african superpopulation, while the right plot shows the same results including african samples" - 'African' is not capitalised (twice).
- p. 45 Extended Data Fig. 11 legend
The left plot excludes the african
Judge's note: Viewed the page 45 image. Lower-case 'african' appears twice in the legend. Capitalisation slip only.
Second check: none (judge only)
Report a problem with this finding
Typo Reporting summary, Sequencing depth: "H3Kme1: Average 288 million reads per sample" - the histone mark should be H3K4me1.
- p. 62 Reporting summary, ChIP-seq, Methodology, Sequencing depth
H3Kme1: Average 288 million reads per sample
Judge's note: Viewed the page 62 image (this page's text is OCR, so the image was checked). The Sequencing depth box does print 'H3Kme1', while the Antibodies and Peak calling boxes on the same page print 'H3K4me1'. The '4' is missing in the page itself, not an OCR error.
Second check: none (judge only)