Enterotypes of the Human Gut Microbiome
2011 · paper · Biomedicine
A correction notice for a Nature paper on enterotypes of the human gut microbiome. It fixes an error in the Supplementary Information, where mean Actinobacteria abundance was listed as 82% instead of 8.2%.
Checked: PDF from media.springernature.com, fetched 4 Oct 2026. Page numbers are those of that PDF.
6 findings
Substance Supplementary Methods 4.3: "Thus we chose 65% and 85% as the thresholds for the genus and phylum level assignments" reverses the pairing; it should be phylum (65%) and genus (85%).
- p. 8 Supplementary Methods 4.3
Thus we chose 65% and 85% as the thresholds for the genus and phylum level
- p. 7 Supplementary Methods 4.2
(>65% for phylum and >85% for genus
- p. 8 Supplementary Methods 4.3, end of paragraph
Thus we chose 65% and 85% as the thresholds for the genus and phylum level assignments.
- p. 7 Suppl. Methods 4.2
>65% for phylum and >85% for genus
Judge's note: The sentence just before on p8 gives 65% for phylum (0.77% FPR) and 85% for genus (1.84% FPR). Methods 4.2 on p7 says ">65% for phylum and >85% for genus", and Notes 2.2 (pp17-18) uses 85% for genus. The concluding sentence lists the ranks in the opposite order.
Second check: another model agreed
Report a problem with this finding
Substance Supplementary Methods 4.3 gives the marker-gene subset as "a subset of 835 genomes" and then "801 of the 853 genomes (94.6%)"; the two counts disagree, and 94.6% matches neither denominator.
- p. 8 Supplementary Methods 4.3
from a subset of 835
- p. 8 Supplementary Methods 4.3
801 of the 853 genomes (94.6%)
- p. 8 Supplementary Methods 4.3
from a subset of 835 genomes
- p. 8 Supplementary Methods 4.3
since 801 of the 853 genomes (94.6%) contained
Only the authors can say what is right here.
Judge's note: Read on the p8 image. Both sentences describe the same redundancy-reduced subset, and nothing in the text explains two different sets. 801/853 = 93.9% and 801/835 = 95.9%, so neither rounds to 94.6%. Only the authors can say which number is wrong.
Second check: another model agreed
Report a problem with this finding
Substance Notes 4.4 reports "leave-one-sample-out cross-validation accuracy of 90.1% (Sanger data set)", but with 33 samples and accuracy defined as correct/all samples, the possible values are k/33 (87.9%, 90.9%), and 90.1% is not one of them.
- p. 21 Supplementary Notes 4.4
validation accuracy of 90.1% (Sanger data set)
- p. 19 Supplementary Notes 4.1
The remaining 33 Sanger-based metagenomic samples were clustered
- p. 21 Supplementary Notes 4.4
validation accuracy of 90.1% (Sanger data set), 86% (16S data set) and 97.6% (Illumina data set)
- p. 12 Supplementary Methods 9
ratio between the number of correctly assigned samples and the number of all samples
Only the authors can say what is right here.
Judge's note: Methods 9 (p12) defines accuracy as the ratio of correctly assigned samples to all samples, under leave-one-sample-out cross-validation with a decision tree. Notes 4.1 gives 33 Sanger samples. The 16S value (132/154 = 85.7%, printed as 86%) and the Illumina value (83/85 = 97.6%) are both reachable; 90.1% is not reachable with 33 samples, nor with 35. 90.9% may have been intended. Only the authors can confirm.
Second check: another model agreed
Report a problem with this finding
Substance Notes 2.2 gives Prevotella across all samples as "0.0%-35.5%", but Notes 5.1.2 gives Prevotella in enterotype 2 as "(5.8%-35.9%)". The maximum within one enterotype is higher than the maximum over all samples.
- p. 18 Supplementary Notes 2.2
Prevotella (mean abundance: 4.4%, standard deviation: 9.5%, 0.0%-35.5%)
- p. 22 Supplementary Notes 5.1.2
Prevotella (5.8%-35.9%)
- p. 18 Notes 2.2
0.0%-35.5%
Only the authors can say what is right here.
Judge's note: Checked page images p18 and p22. Notes 2.2 refers to "at least 32 out of 35 samples", and Notes 4.1 says the 33 clustered samples are what remains after removing two of them, so the enterotype samples are a subset of the overall set. A subset maximum cannot exceed the overall maximum. Both are genus abundances from the same reference-genome mapping, and no different normalisation is stated. Bifidobacterium has the same maximum (20.3%) in both places. Only the authors can say which value is right.
Second check: another model agreed
Report a problem with this finding
Reference or label Supplementary Notes 1.2 ("reference genome mapping approach (See Supplementary Methods Section 4.1)") and Notes 2.2 ("reference genome mapping procedure explained in Supplementary Methods Section 4.1") point to 4.1, but the mapping procedure is in Section 4.2.
- p. 15 Supplementary Notes 1.2
We used the reference genome mapping approach (See Supplementary Methods Section 4.1)
- p. 17 Supplementary Notes 2.2
reference genome mapping procedure explained in Supplementary Methods Section 4.1
- p. 15 Notes 1.2
(See Supplementary Methods Section 4.1)
- p. 17 Notes 2.2
explained in Supplementary Methods Section 4.1
Judge's note: On p7, Section 4.1 is 'Reference genome set', which only describes how the genomes were collected and classified. Section 4.2, 'Reference genome mapping', gives the BLASTN mapping, thresholds and fragment counting. Both pointers were checked on p15 and p17.
Second check: another model agreed
Report a problem with this finding
Reference or label Supplementary Notes 2.2: "85% DNA sequence similarity threshold, which can reliably assign a sequence at the genus level (Supplementary Notes Section 1.2)" points to the section 'Comparing different sequencing technologies'; the thresholds are derived in Supplementary Methods Section 4.3.
- p. 18 Supplementary Notes 2.2
threshold, which can reliably assign a sequence at the genus level (Supplementary Notes Section 1.2)
- p. 18 Notes 2.2
(Supplementary Notes Section 1.2)
- p. 18 Notes 2.2
reliably assign a sequence at the genus level (Supplementary Notes Section 1.2)
- p. 18 Supp. Notes 2.2
which can reliably assign a sequence at the genus level (Supplementary Notes Section 1.2)
Judge's note: The p18 image confirms the pointer. On p15, Notes 1.2 compares Sanger and 454 and says nothing about similarity thresholds. The thresholds are derived in Methods 4.3 on p8. The Section 4.1 pointers raised in this cluster are covered in C011.
Second check: another model agreed
Report a problem with this finding
12 minor (typos and the like; most checked by the judge only)
Wording Correction notice says "see Supplementary Information Table of Contents for details", but the Table of Contents (PDF pp. 2-4) has only section titles and page numbers and no details of the correction.
- p. 1 Correction notice
see Supplementary Information Table of Contents for details
- p. 2 Table of Contents
Table of Contents
Judge's note: Viewed the notice on p1 and all three TOC pages (pp2-4). None of them mentions the Actinobacteria correction. The notice also says only "section 2.2", and that number exists in both the Methods and the Notes. The corrected value (8.2%) does appear in Supplementary Notes 2.2 on p18. Minor wording point about where readers are sent.
Second check: another model agreed
Report a problem with this finding
Wording Notes 4.1 says "The remaining 33 Sanger-based metagenomic samples were clustered", but the sample counts imply the 33 include the 2 American metagenomes "determined using pyrosequencing" (Notes 1).
- p. 19 Supplementary Notes 4.1
The remaining 33 Sanger-based metagenomic samples were clustered
- p. 15 Supplementary Notes 1
the latter include only 2 out of 18 metagenomes from one American study5 determined using pyrosequencing
Only the authors can say what is right here.
Judge's note: Notes 1 (p15) lists 22 European, 13 Japanese and 4 American samples, of which 2 American samples are pyrosequenced (ref 5). Notes 2.2 uses n = 2 to 35 for "all samples except infants". Removing the two American Sanger samples (ref 4) from the 35 leaves 33, and those 33 would include the two 454 samples. So the label 'Sanger-based' does not fit all 33. The authors may use 'Sanger dataset' loosely, so this is a wording query.
Second check: another model agreed
Report a problem with this finding
Wording Notes 5.1.3 gives "Akkermansia (0.0%-8.9%)" in enterotype 3, which is above the maximum for its phylum over all samples, "Verrucomicrobia (... abundance: 0.0 - 8.8%)" (Notes 2.2).
- p. 18 Supplementary Notes 2.2
Verrucomicrobia (mean abundance: 1.3%, standard deviation: 2.1%, abundance: 0.0 - 8.8%)
- p. 22 Supplementary Notes 5.1.3
Akkermansia (0.0%-8.9%)
Only the authors can say what is right here.
Judge's note: Checked p18 and p22. Methods 4.2 (p7–8) says abundances at higher ranks are sums of the genome abundances under them, so a genus should not exceed its phylum in any sample. However, the phylum (65%) and genus (85%) similarity thresholds differ, so the unassigned fraction, and therefore the denominator, can differ slightly between ranks. That could just about account for a 0.1-point excess. The difference is small, so this is a query rather than a definite error.
Second check: another model agreed
Report a problem with this finding
Wording Bacteroides over all samples is given as "0.0%-54.7%" (Notes 2.2), but Enterotype 1 Bacteroides is "between 21.3% and 54.3%" and the other enterotypes reach only 15.3% and 17.6%.
- p. 18 Supplementary Notes 2.2
13.4%, 0.0%-54.7%
- p. 22 Supplementary Notes 5.1.1
between 21.3% and 54.3%
Only the authors can say what is right here.
Judge's note: Checked p18 and p22. The only non-infant samples outside the clustering are the two American Sanger samples, which Notes 4.1 says were removed for very low Bacteroidetes, so they are unlikely to hold the Bacteroides maximum. The overall maximum would then be expected to equal the enterotype 1 maximum. This depends on the overall set being the 35 non-infant samples, which the "32 out of 35 samples" wording suggests but does not state outright. Hence a query of minor weight.
Second check: another model agreed
Report a problem with this finding
Typo Supplementary Methods 2.1: "Shotgun randomly shared DNA libraries" should read "randomly sheared".
- p. 6 Supplementary Methods 2.1
Shotgun randomly shared DNA libraries
Judge's note: The page image confirms the printed word is "shared". Shotgun libraries are made from sheared DNA, so this is a misspelling.
Second check: none (judge only)
Report a problem with this finding
Typo Supplementary Methods 7.5: terms (c) and (d) read "the number of related samples in new clusters but not in the enterotype clusters", while (a) and (b) count "pairs of" samples; the Rand index needs pair counts throughout.
- p. 12 Supplementary Methods 7.5
(c) the number of related samples in new clusters but not in the enterotype clusters
Judge's note: Checked p12. Items (a) and (b) say 'pairs'; items (c) and (d) leave the word out. The intended meaning is clear from the standard Rand index R = (a+b)/(a+b+c+d), so this is a missing word.
Second check: none (judge only)
Report a problem with this finding
Typo Supplementary Methods 13: "Examples with a across SD/within SD ratio >1" should read "an across-SD/within-SD ratio".
- p. 14 Supplementary Methods 13
Examples with a across SD/within SD ratio >1
Judge's note: The p14 image confirms the wording "with a across SD". The article is wrong before a vowel.
Second check: none (judge only)
Report a problem with this finding
Typo Supplementary Notes 2.1: "we counted the number of reads per samples, since this is not used..." (across the page break) is ungrammatical; it should read "per sample".
- p. 16 Supp. Notes 2.1
number of reads per
Judge's note: The sentence runs from p16 to p17, and the images show "reads per" followed by "samples, since". Grammar slip only.
Second check: none (judge only)
Report a problem with this finding
Typo Duplicated word in Notes 4.2: "we believe that that Enterotype 3 is driven".
- p. 20 Supplementary Notes 4.2
we believe that that Enterotype 3 is driven
- p. 20 Supp. Notes 4.2
we believe that that Enterotype 3
Judge's note: The duplicated 'that that' is visible in the crop of p20.
Second check: none (judge only)
Report a problem with this finding
Typo Notes 5.1.3 prints "unclassified Lachnospiraceae (0.6%-8.2)", with no % sign on the upper bound.
- p. 22 Supp. Notes 5.1.3
(0.6%-8.2)
Judge's note: Page image p22 shows "(0.6%-8.2)". All the other ranges on the page have % on both bounds.
Second check: none (judge only)
Report a problem with this finding
Typo Reference 48 has no title: "Garrity, G. M. et al. (Michigan State University Board of Trustees, MI, USA, 2007)."
- p. 24 References, entry 48
Garrity, G. M. et al. (Michigan State University Board of Trustees, MI, USA, 2007).
Judge's note: The crop of p24 shows a blank gap where the title should be. Methods 4.1 cites ref 48 as the "Taxonomic Outline of Bacteria and Archaea", which gives the evident fix.
Second check: none (judge only)
Report a problem with this finding
Typo Reference 54 ends "doi:btp336 [pii]" (no real DOI and no year), and reference 86 reads "doi:btl446 [pii]" before the real DOI.
- p. 24 References, entry 54
doi:btp336 [pii]
- p. 26 References, entry 86
doi:btl446 [pii]
Judge's note: The crops of p24 and p26 confirm both. In ref 86 the real DOI 10.1093/bioinformatics/btl446 and the year follow on the next line. Ref 54 ends at the pii, with no DOI and no year.
Second check: none (judge only)