A Global Reference for Human Genetic Variation

• p. 5 Figure 4a legend

  LWK

• p. 19 Extended Data Figure 9a, panel titles

  YRI

Only the authors can say what is right here.

Judge's note: Read Fig. 4a (p.5, full page and enlarged crops) and Extended Data Fig. 9a (p.19). The colour reversal is plain. In Fig. 4a the yellow curve has the lowest plateau (about 0.94) and the brown one about 0.96; in Extended Data Fig. 9a the yellow 'LWK' panel likewise has the lower plateau and the brown 'YRI' panel the higher, which is consistent with one figure's labels being swapped rather than only its colours, but values read from graphs this close cannot settle it. Elsewhere (Figs 3a, 4c, Extended Data Fig. 6) LWK is brown and YRI pale orange. Only the authors can say whether labels or merely colours differ.

Second check: another model agreed

• p. 1 Data set overview, second paragraph

  depth 5 473) for 427 individuals

• p. 28 Section 3.5.2

  Average coverage of the CG sequencing was ∼55×

• p. 1 Data set overview, 2nd paragraph

  deep Complete Genomics data (mean depth = 47×) for 427 individuals

• p. 28 Section 3.5.2 CG data submission, 'Average coverage' sentence

  Average coverage of the CG sequencing was ∼55×.

Only the authors can say what is right here.

Judge's note: Read the page image of main p.1 (text layer garbles '= 47×' as '5 473') and the supplement crop. Supplement 3.5.1 says Complete Genomics supplied deep data for 427 samples, so both describe the same data set. Searched the supplement text for a definition distinguishing the two (mapped vs gross coverage, subset); none found. Could be two legitimate measures, so this is a query, not a correction.

Second check: another model agreed

• p. 1 Introduction, second paragraph

  (meandepth 5 7.43)

• p. 4 Table 1, row 'Mean coverage'

  8.2

• p. 1 Introduction, 2nd paragraph

  whole-genome sequencing (mean depth = 7.4×)

• p. 4 Table 1, row 'Mean coverage', EUR column

  7.4

Only the authors can say what is right here.

Judge's note: Read main p.1 and Table 1 on p.4 from the page images. Sample counts 661/347/504/503/489 sum to 2,504. The weighted mean of the table row is 7.82, and no group is below 7.4, so the overall mean cannot be 7.4 on the same basis. Searched article and supplement for a definition of either figure (autosomal only, mapped vs raw, 2,535 sequenced vs 2,504 released); the supplement gives per-centre coverage only and no explanation. Different bases are possible, so this is a query.

Second check: another model agreed

• p. 3 Sharing of genetic variants, f2 paragraph

  296 generations ago (7,410 to 8,892 years

• p. 3 Same paragraph

  143 generations ago (3,570 to 4,284 years ago)

• p. 4 Figure 2b top axis title

  20–30 years per generation

• p. 16 Extended Data Figure 6d, top axis (Years) against bottom axis (Generations 5000)

  100k-150k

Only the authors can say what is right here.

Judge's note: Read p.3 text, Fig. 2b axis title on p.4 and Extended Data Fig. 6d on p.16 from the images. The ratios of the printed year bounds to the generation counts are 25.0 and 30.0 for both estimates; the cited figure's top axis gives 20 and 30. Searched article and supplement text (pages in scope) for a stated generation time for the f2 analysis; none found. The estimates cite ref. 13 and may follow that method's convention, so this is a query about an unstated, inconsistent conversion.

Second check: another model agreed

• p. 3 Sharing of haplotypes and imputation

  and SNPs overlap, but slightly reduced for multi-allelic indels

• p. 3 Extended Data Figure 9b legend

  Multi−allelic Indels

Only the authors can say what is right here.

Judge's note: Read the sentence on p.3 and Extended Data Fig. 9b with its legend on p.19 (the figure is on PDF page 19, not page 3). No curve or legend entry for SNP/indel overlap sites, and the legend text ('Performance of imputation from phase 3 by variant class') does not say that class is folded into another. Minor mismatch between text and cited figure.

Second check: another model agreed

• p. 3 Sharing of haplotypes and imputation

  was >95% for common variants in each population

• p. 5 Figure 4a, yellow curve (YRI)

  YRI

Only the authors can say what is right here.

Judge's note: Read Fig. 4a on p.5 at full page and in an enlarged crop. Measuring against the 0.8 and 1.0 ticks, the yellow (YRI) plateau sits at about 0.94, slightly below 0.95, while the other five curves plateau at about 0.955 to 0.975. This is a reading from a graph with an uncertainty of about 0.01, so it is a query on the wording 'each population' rather than a firm error.

Second check: another model agreed

• p. 12 Extended Data Fig. 2 caption, panel a

  The power of discovery within the main data set for SNPs and indels identified within an overlapping sample of 284 genomes

• p. 12 Extended Data Fig. 2a legend

  CG.SV

Judge's note: Viewed the full page image. The legend of panels A and B lists CG.INDEL, CG.SNP, CG.SV and HRC.SNP, and a CG.SV curve is drawn. The caption for a (and b) does not mention SVs; only the caption for c does. The legend itself identifies the curve, so this is a caption incompleteness, not a data error.

Second check: another model agreed

• p. 27 table in 3.5.1, rows KHV and PUR and last column header

  Overlap with

• p. 27 table row labels

  KHV*

Judge's note: Viewed supp p27: the table has no footnote and section 3.5.2 starts directly below it. The text on p26 partly explains both: the PUR and KHV trios were sequenced from both blood and cell-line DNA (which accounts for those rows and the Total 433 against Unique 427), and "287 samples overlapped with samples in the primary low-coverage and exome data, while 284 overlapped with the samples in the final release" matches "287 (284)". The reader can infer the meaning, but the asterisks themselves are never explained.

Second check: another model agreed

• p. 80 7.1.1 last paragraph

  differed by two log units

Judge's note: Viewed supp-01 page 80, last paragraph. The two sentences are printed as quoted. Read literally, a difference of exactly two log units meets both the calling condition and the no-call condition; the second sentence makes the intended rule clear (call only when the difference exceeds two), so this is loose wording with no effect on any reported number. Searched the supplement text for another statement of the threshold and found none.

Second check: another model agreed

• p. 1 Introduction paragraph 2

  first-degree relatives

• p. 3 Box 1

  first degree relatives

Judge's note: All three occurrences read in the page images of p.1 and p.3. Cosmetic inconsistency only.

Second check: none (judge only)

• p. 3 Box 1, step (3) and figure labels

  The biallelic and multi-allelic haplotypes

• p. 3 Box 1, step (2a)

  high-confidence bi-allelic variants

Judge's note: Both spellings read in Box 1 on the p.3 page image; 'bi-allelic' also on p.1. Cosmetic inconsistency.

Second check: none (judge only)

• p. 5 Figure 4 legend, panel a

  The insert compares

• p. 3 Sharing of haplotypes and imputation

  (Fig. 4a, inset)

Judge's note: Legend on p.5 reads 'insert'; text on p.3 reads 'inset'. The small plot within panel a is an inset. Cosmetic.

Second check: none (judge only)

• p. 80 Heading 7.1.1

  Filtered SNP callset and phylogentic tree

• p. 3 Table of contents, 7.1.1

  phylogentic tree

• p. 80 7.1.1 heading

  phylogentic tree

• p. 80 7.1.1 first sentence

  extracted from from the union

Judge's note: Viewed supp-01 page 80 and the contents crop on page 3. Both the heading and the contents entry print "phylogentic"; the body text just below spells "phylogenetic" correctly. The doubled "from from" that one reviewer bundled in here is handled in C077.

Second check: none (judge only)

• p. 5 Body text under Figure 3, left column

  we find 3–6 tagging variants with 100 kb of variants

• p. 5 Resolution of genetic association studies, p. 72, left column

  3–6 tagging variants with 100 kb of variants

Judge's note: Read in the page image and crop of p.5: printed as 'with 100 kb of variants'. Fig. 4b plots tagging variants against distance up to 100 kb, so 'within' is the evident intended word.

Second check: none (judge only)

• p. 71 Section 5.7, first line of page

  eludcidate

• p. 6 Discussion, 1st paragraph

  while the remainder were removed by our filter-

• p. 6 Discussion, 1st paragraph

  1.6 million had frequency <0.5%

• p. 70 Section 5.7

  340K of them failed our

Judge's note: All five slips read in the full-page images of supp pp. 70 and 71 and printed as quoted. One reviewer note in this cluster raises a different point: that the article's Discussion (main p. 6, read in the crop: "1.6 million had frequency <0.5% and may be missing from our current read set, while the remainder were removed by our filtering processes") says more than section 5.7, which attributes 340K SNPs to the SVM filter, 57K to overlap with larger variants, 84K to pipeline losses, and does not give a 1.6 million figure. That is not confirmed as a defect here: the Discussion sentence is a one-line summary, section 5.7 points to further detail on the FTP site, which is not among the files, and "filtering processes" can reasonably cover the pipeline as a whole. The verdict and severity apply to the typographical slips only.

Second check: none (judge only)

• p. 7 Acknowledgements, funding

  WT0855322/Z/08/Z (R.L.)

• p. 7 Acknowledgements, funding

  WT0855322/Z/08/Z

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that Wellcome Trust grant references have six digits before the '/Z' suffix, and what the correct number is. The crop of p.7 shows 'WT0855322/Z/08/Z' as quoted, and every other reference of this form in the list has six digits, so a stray digit is likely, but the correct number cannot be determined from the files.

Second check: none (judge only)

• p. 7 Acknowledgements, funding

  089276/Z.09/Z (S.D.)

• p. 7 Acknowledgements, funding

  089276/Z.09/Z

Judge's note: Crop C016-a shows '089276/Z.09/Z' with a full stop clearly visible; every other Wellcome reference in the same sentence uses the NNNNNN/Z/YY/Z form. Typographic slip.

Second check: none (judge only)

• p. 7 Acknowledgements, NIH grants

  P01GM99568 (S.R.B.), R01GM59290

• p. 7 Acknowledgements, NIH grants, two lines later

  T32GM7790 (C.D.B., A.R.M.), P01GM99568 (S.R.B.)

• p. 7 Acknowledgements, funding

  P01GM99568 (S.R.B.)

Judge's note: Both occurrences are visible on the page image two lines apart, with the same grant number and the same initials. No reason for the repeat is given; a duplicated entry.

Second check: none (judge only)

• p. 8 Analysis group, Boston College

  Jiantao Wu39

• p. 8 Structural variation group, Boston College

  Jiantao Wu23

Judge's note: Both superscripts read from the page 8 image. The same author under the same institution heading carries two different affiliation numbers in two groups. Either could be a legitimate address (present versus former), so this is a consistency point rather than a definite error.

Second check: none (judge only)

• p. 8 Analysis group, Boston College

  Mengyao Zhang40

• p. 8 Structural variation group, Boston College

  Mengyao Zhang23

Judge's note: Both superscripts read from the page 8 image; affiliation 40 read in the text layer (Harvard Medical School). Same author, same heading, two affiliation numbers. Either may be a legitimate address; consistency point only.

Second check: none (judge only)

• p. 8 Analysis group, Boston College

  Leong38

• p. 9 Exome group, Boston College

  Wen Fung Leong23

Judge's note: Page 8 image shows 'Wen Fung Leong38' in the Analysis group; page 9 image shows 'Wen Fung Leong23' in the Exome group; affiliation 38 read on the page 9 image. Same author and heading, two affiliation numbers; either may be legitimate, so a consistency point.

Second check: none (judge only)

• p. 9 Author list, Structural variation group, University of Texas MD Anderson Cancer Center

  Ken Chen121 (Principle Investigator)

• p. 9 Structural variation group, University of Texas MD Anderson Cancer Center

  Principle Investigator

Judge's note: Crop C021-a shows 'Principle Investigator' for Ken Chen, while the neighbouring entries on the same lines read 'Principal Investigator'. Misspelling.

Second check: none (judge only)

• p. 9 Author list, Functional interpretation group, Yale University

  Mark B. Gerstein113,1114,115

• p. 9 Functional interpretation group, Yale University

  Gerstein113,1114,115

• p. 9

  Mark B. Gerstein113,1114,115 (Principal Investigator) (Co-Chair)

Judge's note: Crop C022-a shows the superscript '113,1114,115' on the page; crops C018-b, C020-b and C021-a show '113,114,115' for the same author in the Analysis, Structural variation and Exome groups. No affiliation 1114 exists (the list ends at 168, seen in C028-a).

Second check: none (judge only)

• p. 9 Affiliation list, no. 32

  Heath Genomics, UNU-MERIT

Judge's note: Crop C023-a shows 'Public Heath Genomics' on the page image. Plain misspelling of 'Health'.

Second check: none (judge only)

• p. 10 Affiliation list, no. 60

  60General Hospital and Harvard

Judge's note: Crop C025-a shows affiliation 60 beginning 'General Hospital and Harvard Medical School, Boston, Massachusetts 02114', and in the same crop affiliations 53 and 123 give 'Massachusetts General Hospital' at the same city and postcode. The hospital's name is truncated.

Second check: none (judge only)

• p. 10 Affiliation list, no. 91

  and Bioinfomatics, University of Michigan

Judge's note: Crop C026-a shows 'Bioinfomatics' on the page image (the 'r' is missing). Other affiliations on the same page spell 'Bioinformatics' correctly.

Second check: none (judge only)

• p. 10 Affiliation list, no. 98

  delgi Studi di Sassari

Judge's note: Crop C027-a shows 'delgi Studi di Sassari' on the page image. 'delgi' is not an Italian word; 'degli' is the article that the standard form 'Università degli Studi di ...' takes. Spelling slip.

Second check: none (judge only)

• p. 10 Affiliation list, no. 101

  University of Chapel Hill, North Carolina 27599

Judge's note: Crop C028-a shows 'University of Chapel Hill, North Carolina 27599' in affiliation 101. The text of the author list (page 8/9) places Yun Li101 under the heading 'University of North Carolina at Chapel Hill', and affiliation 102 writes 'University of North Carolina at Charlotte' in full. 'North Carolina at' has been dropped from the institution's name.

Second check: none (judge only)

• p. 11 Section 2.2

  RPMI 1650

• p. 11 2.2 Lymphoblastoid cell line establishment

  RPMI 1650 with

• p. 11 2.2

  phytohemaglutinin

• p. 11

  incubating with Epstein Barr virus and phytohemaglutinin in RPMI 1650

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: the standard culture medium is named RPMI 1640 and no RPMI 1650 is known; the paper itself does not give the name elsewhere. The crop shows "phytohemaglutinin in RPMI 1650" exactly as quoted. The spelling of phytohemagglutinin (two g's) is a plain misspelling; the medium number needs the authors to confirm.

Second check: none (judge only)

• p. 12 Section 2.2, second paragraph

  and identify QCs

• p. 12 2.2, second paragraph

  identify QCs

• p. 12 2.2, second paragraph

  ITU,MSL

• p. 12

  sterility and identify QCs

Judge's note: Viewed the crop: the text reads "sterility and identify QCs. Identity by state (IBS) data analysis..." and "ESN, GWD, ITU,MSL, PJL and STU". The preceding paragraph is about confirming sample identity, so "identity" is the intended word.

Second check: none (judge only)

• p. 17 Section 3.2.3, first paragraph

  f) We then performed an agarose gel

• p. 17 3.2.3 BGI, library construction steps

  f) We then performed an agarose gel electrophoresis

• p. 17 3.2.3 BGI

  Illumina Hiseq

Judge's note: Viewed supp p17. The lettering runs a, b, c, d, f, g. Whether a step was dropped or the letters are simply off cannot be told from the page. "Hiseq" appears once here against "HiSeq" in the next paragraph and throughout the supplement.

Second check: none (judge only)

• p. 23 3.4.1 Fingerprint concordance

  then an reassignment of the sample genotype

Judge's note: Viewed the crop; the text reads as quoted.

Second check: none (judge only)

• p. 25 Section 3.4.2, Post-Amplification

  Resupension, Hybrisation

• p. 25 3.4.2

  Genotypying pipeline

• p. 25 3.4.2 Post-Amplification

  Resupension, Hybrisation to beadchip

• p. 25 3.4.2 Post-Amplification

  Illumina Autloader 2.Xs

Judge's note: Viewed supp p25; all four words are printed as quoted.

Second check: none (judge only)

• p. 30 3.6.2

  to used that latest software

• p. 31 3.6.2

  created using using ‘bwa sampe’

• p. 32 3.6.2

  to the include both

• p. 31

  were created using using ‘bwa sampe’ or ‘samse’

Judge's note: Viewed the three crops on supp pp30, 31 and 32; each reads as quoted, and the repeated "using using" is on one line.

Second check: code check

• p. 36 Section 4, first paragraph

  from the the Project

• p. 36 4 Variant calling, first paragraph

  a range additional structural variants

• p. 36

  single set of haplotypes from the the Project

• p. 36

  single set of haplotypes from the the Project. Details of

Judge's note: Viewed both crops of supp p36; both phrases read as quoted, and "the the" is on one line.

Second check: code check

• p. 37 4.1.1 Baylor SNPTools

  of the SNPs called the

Judge's note: Read on the page image of supplement p. 37; the sentence is printed as quoted.

Second check: none (judge only)

• p. 38 Section 4.1.2

  --contimation-estimates

• p. 64 Section 5.2.2

  --contamination-estimates

• p. 38

  via the --contimation-estimates flag

Judge's note: Misspelling read on the page image of p. 38 (monospaced flag name). The correct spelling on p. 64 was checked in the extracted text only.

Second check: none (judge only)

• p. 40 Section 4.1.4, item 3

  has the he maximum

• p. 40 4.1.4 (3)

  the he maximum

• p. 40 4.1.4 (1) item 1

  until there a base

• p. 42 4.1.4

  (complex enough and without cycles we try

Judge's note: All five read on the page images of pp. 40 and 42, printed as quoted.

Second check: none (judge only)

• p. 42 4.1.4 (2)

  across the ER

• p. 42

  with the reference sequence across the ER

Judge's note: Read on p. 42. "AR" is defined in the heading "(0) Defining the ActiveRegions (AR)" on p. 38 and used on pp. 40 and 42. A search of the supplement text finds no other "ER" and no definition of it.

Second check: none (judge only)

• p. 50 Section 4.1.9, Pipeline and filtering

  descibed

• p. 50 4.1.9 Pipeline and filtering

  descibed here

• p. 50

  using SGA version 0.10.8 using the pipeline descibed here

Judge's note: Read on the page image of p. 50.

Second check: none (judge only)

• p. 51 Section 4.1.10

  details about about our analysis

• p. 51 4.1.10

  also preforms pedigree-

• p. 51 4.1.10

  More details about about our analysis

• p. 51

  rtgVariant also preforms pedigree-

Judge's note: Both read on the page image of p. 51. The repeated "about" straddles a line break ("More" ends one line, "details about about" begins the next) but both words are printed.

Second check: code check

• p. 52 4.2 Micro-satellites

  due allelic dropout

• p. 52 4.1.10

  was 2.06 The false-positive rate

• p. 52

  The Ts/Tv ratio for these sites was 2.06 The false-positive

Judge's note: Both read on the page image of p. 52.

Second check: none (judge only)

• p. 57 Section 4.3.5

  hybrdization

• p. 57 4.3.5 Read-Depth (dCGH)

  hybrdization (dCGH)

• p. 57 4.3.4 CNVnator

  to accounts for multiple hypothesis testing

• p. 57

  digital comparative genomic hybrdization (dCGH)

Judge's note: Both read on the page image of p. 57.

Second check: none (judge only)

• p. 59 Section 4.3.8

  less than than 90%

• p. 59 4.3.7 Pindel (note)

  where is was

• p. 59 4.3.8

  less than than

• p. 59

  Several samples with less than than 90% properly mapped read pairs

Judge's note: Both read on the page image of p. 59.

Second check: code check

• p. 60 Section 5, first paragraph

  to create and final integrated callset

• p. 60 5, list item 1

  As described Section 5.1

• p. 60 5, list item 1

  passed an Support Vector Machine

Judge's note: All read on the page image of p. 60. The missing "in" occurs in both list items 1 and 2.

Second check: none (judge only)

• p. 68 Section 5.5.2

  mono-monomorphic reference sites

• p. 60 Section 5 first paragraph

  to create and final integrated callset

• p. 68 5.5.2

  mono-monomorphic

• p. 68

  All post-phasing mono-monomorphic reference sites were excluded

Judge's note: Both read in images (supp p. 68 full page; crop of p. 60) and printed as quoted. Copy-editing slips.

Second check: none (judge only)

• p. 63 5.1.3 Filtering of biallelic SNPs

  Next, we based on this annotations

Judge's note: Read in the crop of supp p. 63; the sentence is printed as quoted. Evident fix: "Next, based on these annotations, we evaluated...".

Second check: none (judge only)

• p. 64 Section 5.2.2

  Were the alignment to the graph had

• p. 64 5.2.3

  prior to perform

Judge's note: Both passages read in the full-page image of supp p. 64 and are printed as quoted. Copy-editing slips only.

Second check: none (judge only)

• p. 72 Section 6.1, third paragraph on page

  in order to achieve a FDR below

• p. 72 6.1

  An allele was considered as validated it was supported

• p. 72 6.1, paragraph after weighted FDR

  achieve a FDR below

Judge's note: Read the full-page image of supp p. 72: "FDR below" ends one line and "while maintaining" starts the next with no value, so this is not an extraction loss. The surrounding text gives the requirement as 5% ("met the 5% FDR requirement"), so the fix is evident. "as described elsewhere" in the previous sentence carries no reference, which is a style point and not counted.

Second check: none (judge only)

• p. 73 6.2.1 EMBL PCR validation

  PCR was preformed using

• p. 73 6.2.1

  send for capillary sequencing

• p. 73 6.2.1

  preformed

• p. 73

  PCR was preformed using 10 ng of

Judge's note: Both read in the full-page image of supp p. 73 and printed as quoted.

Second check: none (judge only)

• p. 75 Section 6.2.4

  82.3% (28/34)

Judge's note: Read in the crop of supp p. 75. 28/34 = 82.353%, not a rounding tie, so conventional rounding gives 82.4%; 82.3% is the truncated value. The neighbouring figures are rounded conventionally and check out (111/113 = 98.2%; 28/32 = 87.5%). A difference of 0.1 percentage point with the fraction printed beside it; cosmetic.

Second check: code check

• p. 76 6.3.1 Data Production

  loaded and ran on a pulse field gel

Judge's note: Read in the crop of supp p. 76. "was loaded and ran" is a grammar slip for "run". "pulse field gel" for "pulsed-field gel" is a common laboratory spelling and is not counted.

Second check: none (judge only)

• p. 80 Section 7.1.1, first sentence

  extracted from from the union

• p. 80

  The set of biallelic SNPs was extracted from from the union genotype likelihoods.

• p. 80

  biallelic SNPs was extracted from from the union genotype

Judge's note: Viewed supp-01 page 80. The first sentence under heading 7.1.1 prints "extracted from from the union" on a single line, so it is not a line-break or extraction artefact.

Second check: code check